1. Introduction
Smartwatches have evolved into advanced health monitoring tools, enabling the continuous collection of important biometric data from users. In Ref. [1], modern wearable devices are equipped with various sensors that record vital signs continuously, enhancing self-monitoring and the early detection of health issues. For instance, a smartwatch can continuously monitor a user’s heart rate and physical activity levels, track sleep patterns, and, in some advanced models, perform electrocardiograms (ECGs) or measure blood pressure. The purpose of Dhar et al.’s study was to showcase how such devices enhance the self-management of health by providing early indicators of potential issues, such as cardiac arrhythmias or excessive stress. Their findings emphasize significant benefits: continuous monitoring via a smartwatch can motivate users toward healthier habits and contribute to the early detection of health anomalies.
At the same time, the essential limitations of these technologies are also pointed out. For example, the reliability of measurements may be affected by factors such as skin type or wrist movement, and the limited battery life often hinders uninterrupted data collection. Privacy is another major concern, as the sensitive health data collected requires secure management and careful interpretation. Overall, the analysis concludes that while a smartwatch is a valuable complement to health monitoring, its effective use depends on trustworthy technologies and responsible user behavior [1].
With these advantages and drawbacks as a backdrop, a critical next question arises: “How willing are people to adopt smartwatches for health-related purposes, and what factors influence this decision”?
This question was explored by Papa et al. [2], who conducted an empirical study aimed at identifying the determinants of public acceptance and the use of wearable health devices. Their research focused on adults aged 25–40 in India (273 participants), assessing users’ perceptions of device intrusiveness and comfort, as well as perceived usefulness and ease of use. Using a structural equation modeling approach (PLS-SEM), Papa et al. examined how these factors relate to users’ attitudes toward smartwatches and their intention to use them for personal health monitoring. Their findings revealed a nuanced effect: comfort and perceived intrusiveness do not directly affect the intention to adopt the device. Instead, these factors indirectly shape the perceptions of usefulness and usability. Specifically, when a device is ergonomic and not perceived as intrusive in daily life, users are more likely to find it useful and easy to use—key motivators for adoption according to classic technology acceptance models. Indeed, their analysis confirms that high perceived usefulness and ease of use are associated with positive attitudes and strong usage intentions. They conclude that improvements in the design of smartwatches—making them less intrusive and more comfortable—can enhance their perceived value and thus encourage more people to integrate them into their health monitoring routines.
However, as the acceptance of smartwatches grows and they become part of everyday life, a new question emerges as follows: “How effectively can these devices be used for specific medical purposes, such as stress management”?
This aspect was examined by Dai et al. [3], who conducted a study focused on the ability of smartwatches to detect and predict stress levels using machine learning techniques (Figure 1). The aim of their research was to compare different stress prediction models using both objective indicators (predefined stress-inducing stimuli) and subjective participant self-reports.
From a clinical standpoint, smartwatches have not fully reached the diagnostic reliability of medical-grade devices but offer promising complementary utility. Some devices, such as the Apple Watch, have received FDA clearance for atrial fibrillation detection, yet many others operate under general wellness classifications. Moreover, privacy concerns regarding continuous biometric data collection remain a key ethical consideration.
The aim of their research was to compare different stress prediction models using both objective indicators (predefined stress-inducing stimuli) and subjective participant self-reports. To achieve this, they supervised a controlled experiment with 32 healthy volunteers, each performing a series of stress-inducing tasks—social, cognitive, and physical—while wearing a commercially available smartwatch equipped with appropriate sensors. Throughout the tasks, the watch continuously recorded physiological signals (e.g., heart rate), and participants were periodically asked to report their perceived stress levels.
This methodology allowed the researchers to develop two types of AI models: one using the objective stress events as ground truth, assuming participants were under stress during tasks, and another using participants’ subjective reports to label stress episodes.
The results are particularly encouraging regarding the effectiveness of smartwatches in stress detection and prediction. The best-performing algorithm—a Support Vector Machine (SVM) classifier trained on objective stress-inducing events—achieved an Area Under the ROC Curve (AUROC) of approximately 0.79, indicating a strong ability to distinguish between stress and calm states. Even when models were based on subjective self-reports of stress, performance remained solid, with AUROC values around 0.72, which is slightly lower due to the variability introduced by individual perceptions of stress.
A noteworthy finding is that tailoring models to individual differences—by calibrating personal thresholds for what constitutes “stress”—further improved prediction accuracy. With this personalized approach, performance reached AUROC levels (0.75) comparable to the objective models. Dai et al. emphasize that the accuracy achieved using a commercial smartwatch is comparable to results from more controlled, laboratory-based studies, demonstrating the potential of consumer-grade wearables in serious health applications [3]. However, they also recognize the lower reliability of models based on subjective reports, underlining the need for further research into how to best collect reliable “ground truth” stress data in real-world settings.
In summary, these recent scientific studies highlight the complementary dimensions of the role smartwatches play in health monitoring, painting a comprehensive picture of both their potential and their challenges. The review by Dhar et al. outlines the broad functionality of these devices, providing the context of benefits and limitations [1]. The work of Papa et al. confirms that technological value only translates into real benefit when users accept and use the devices—certain aspects are enhanced when smartwatches are viewed as both useful and user-friendly [2]. Finally, the study by Dai et al. demonstrates that under suitable conditions, smartwatches can provide reliable measurements for critical health indicators such as stress, achieving results comparable to specialized medical equipment [3].
A common thread across all studies is that smartwatches have the potential to significantly contribute to health promotion: They offer continuous data that supports early diagnosis, encourage users to adopt healthier behaviors, and allow for more personalized wellness monitoring. At the same time, there is a clear need for further advancement—both on a technical level (to improve sensor and algorithm accuracy) and in terms of user design and education—so that smartwatches can be safely and reliably integrated into the healthcare ecosystem for maximum benefit [1,2,3].
Building on the strengths and limitations identified in recent studies, this systematic review examines how smartwatches are currently used to monitor stress and support mental health. To structure this review, we address the following three research questions: Which biometric signals and smartwatch-based sensing technologies are commonly used to detect stress and monitor mental health? What machine learning methods have been applied to process smartwatch data, and how effective are these approaches? What practical limitations and ethical concerns have been reported in the real-world use of smartwatch-based mental health systems?
2. Method
This systematic review was conducted according to the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). A comprehensive literature search was carried out across four electronic databases: PubMed, Scopus, Web of Science, and IEEE Xplore. The search was designed to identify studies related to smartwatches in stress detection and mental health monitoring using relevant search terms and Boolean operators. Specifically, terms such as “smartwatch” in combination with “stress” and “mental health” were used (e.g., “smartwatch” AND “stress” OR “mental health”). These terms were applied to titles, abstracts, and keywords according to the requirements of each database. The search was initially conducted in early 2025 and updated until May 2025 to include the most recent studies. During the search, filters were applied to limit results to English-language publications, studies published within the period 2016–2025, and peer-reviewed journals.
2.1. Inclusion and Exclusion Criteria
We included studies that met five predefined criteria. First, they had to be peer-reviewed journal articles published in English with full-text access. Second, they had to be published between January 2016 and April 2025. Third, they had to examine the use of smartwatches or equivalent wearable devices for stress detection or mental health monitoring, including conditions such as anxiety or depression. Fourth, the study had to present original research—such as clinical trials, user studies, or experimental designs—with physiological data collected from smartwatches (e.g., heart rate, heart rate variability, or electrodermal activity). Fifth, studies had to use algorithmic methods (e.g., machine learning or signal processing) to interpret the data.
We excluded studies that failed to meet the inclusion criteria upon full-text review. Specifically, we excluded non-English articles, non-peer-reviewed sources (e.g., editorials, newsletters, theses), and literature reviews or meta-analyses without original data. We also excluded studies focusing solely on smartphones, fitness bands, or other devices that lacked smartwatch-specific components, as these differ in sensing capabilities and usage context. Furthermore, we excluded articles unrelated to mental health outcomes—for example, those focusing solely on physical activity—and those lacking empirical results or a quantitative evaluation of stress or mental health prediction models.
A study was considered relevant if it used smartwatches as the primary source of biometric data for stress or mental health monitoring and applied algorithmic techniques to analyze the collected signals.
2.2. Study Selection Process
During the study selection process, we adhered to the PRISMA guidelines to ensure transparency. The results of each screening stage are summarized. In Figure 2, a PRISMA flow diagram is presented, summarizing the study identification and selection process. At the end of the process, the literature search yielded 61 studies that met all inclusion criteria and were included in the qualitative synthesis of this systematic review. These studies form the basis of our analysis on how smartwatches are used for stress detection and mental health monitoring in the current literature. Figure 2 illustrates the study identification and selection process; the resulting 61 studies formed the basis for synthesis.
2.3. Quality Assessment of Included Studies
To assess the methodological quality of the included studies, we conducted a qualitative appraisal based on key domains commonly reported in health informatics reviews. These included the clarity of inclusion criteria, sample size adequacy, transparency in sensor data collection, and reporting of algorithm performance metrics (e.g., accuracy, AUC). Given the heterogeneity of study designs and outcomes, we did not perform a formal meta-analysis or apply a standardized risk of bias tool such as ROBINS-I. However, each study was independently reviewed by two authors to identify major methodological flaws or limitations. Discrepancies were resolved through discussion.
In general, most studies clearly reported their sensor setup and algorithm design but varied in terms of participant demographics, data preprocessing techniques, and validation protocols. We highlight these differences in Table 1 and further elaborate on their implications in Section 4.
3. Smartwatches as Stress Assessment Tools
According to Chuah [12], their study is grounded in a multidimensional theoretical framework that combines the analysis of perceived benefits and risks with the monitoring of pre-existing lifestyle imbalances. Within this framework, the methodology relies on Likert-scale questionnaires designed to measure both objective and subjective parameters that determine the intention to continue using a smartwatch. Through analytical techniques, this study identifies how perceived benefits—such as increased productivity, entertainment, and enhancement of social image—activate psychological mechanisms of inspiration and well-being, which in turn positively influence the intention to continue using the device. This approach is characterized by a clear delineation of variables and the application of sequential mediation models that explain how internal psychological processes mediate the effect of perceived benefits on the consumer’s final decision [12].
In addition, Reeder and David’s [13] systematic review closely examines the uses of smartwatches to promote health and wellness, highlighting both their capabilities and limitations. Their methodology involves a rigorous database search using specific selection criteria, and findings are categorized based on device type (e.g., consumer-grade, developer devices, or prototypes), study goals, and application environments. The authors point out that although most studies focus on physical activity monitoring, the real value of smartwatches lies in their ability to support comprehensive health interventions—such as the self-management of chronic conditions, support for therapeutic programs, and continuous vital sign data collection. In addition, the review emphasizes technical and methodological challenges, such as the need for large-scale field studies, data reliability, and the integration of smartwatches with other technologies (e.g., Internet of Things), which are critical for their wider adoption in clinical settings.
The Pomodoro technique, a popular strategy to avoid procrastination, is applied through the Foqus app by breaking tasks into 25 min sessions followed by 5 min breaks. This structure helps users maintain focus and avoid the fatigue that often accompanies long, uninterrupted work periods. The third feature is the use of positive reinforcement messages that support the user’s mental state. The app displays encouraging phrases like “Well done!” after completing a task or session, providing psychological support. These messages enhance users’ self-esteem and sense of achievement. William et al. [14] assessed smartwatch features (e.g., BPM, HRM, Move Alert) and their influence on users’ health. Based on a TAM (technology acceptance model) framework, the study found that while BPM, HRM, and activity alerts positively influenced health behaviors, features related to stress, sleep, and hydration monitoring had limited impact. The authors called for more robust evaluation and feature development in these areas. In addition, Long et al. [15] explored the strengths and weaknesses of smartwatches in health monitoring. They highlighted their usefulness in cardiological data tracking and preventive care, though concerns remain about measurement accuracy and comprehensive health assessments. Varghese et al. [16] studied Parkinson’s disease (PD) and other movement disorders using Apple smartwatches compared to a high-precision shaker. Using ML algorithms (Multilayer Perceptrons), they achieved 74.1% accuracy, 86.5% precision, and 90.5% recall for tremor classification. In addition, Majumder and Deen [17] explored smartphone sensors (e.g., camera, GPS, accelerometers) for health tracking, demonstrating their value in the remote, cost-effective monitoring of cardiovascular, respiratory, and sleep conditions. At the same time, Torrado et al. [18] developed a smartwatch system for emotional self-regulation in autistic individuals. The device used physiological signals and motion tracking to detect meltdowns and support personalized coping strategies. It successfully helped participants recover from episodic crises.
3.1. Stress Management
This aspect was examined by Dai et al. [3], who conducted a study focused on the ability of smartwatches to detect and predict stress levels using machine learning techniques. The aim of their research was to compare different stress prediction models using both objective indicators (predefined stress-inducing stimuli) and subjective participant self-reports. To achieve this, they supervised a controlled experiment with 32 healthy volunteers, each performing a series of stress-inducing tasks—social, cognitive, and physical tasks—while wearing a commercially available smartwatch equipped with appropriate sensors. Throughout the tasks, the watch continuously recorded physiological signals (e.g., heart rate) while the participants wore a commercially available smartwatch equipped with appropriate sensors. The watch continuously recorded physiological signals (e.g., heart rate). Figure 3 illustrates the mechanism by which these signals were acquired, particularly through the use of photoplethysmography (PPG) sensors and optical components embedded in the smartwatch. Participants were periodically asked to report their perceived stress levels.
Furthermore, according to the study by Ding et al. [19], the research focuses on technological innovation through the development of an integrated smartwatch that incorporates sensors for detecting mental stress. In this study, the researchers adopt an approach that combines biochemical measurements (via a cortisol sensor that detects sweat) and physiological parameters (via heart rate variability—HRV). The method involves detailed experimental procedures, such as theoretical calculations (DFT) to estimate the electrostatic potential between the polymer monomer and cortisol, and the design of a sensor based on organic electrochemical transistors (OECTs). This process demonstrates that the sensor offers high sensitivity and reusability, with the ability to convert small changes in cortisol levels into corresponding changes in electric current, enabling the direct and non-invasive monitoring of mental stress [19]. The detailed analyses of the sensor’s various parameters, such as the role of Prussian Blue in enhancing conductivity, exemplify the integration of theoretical and practical approaches in achieving a comprehensive solution. From a broader perspective, these studies present complementary approaches to the evaluation and implementation of smartwatches in healthcare. While Chuah’s [12] study offers a theoretical and psychologically oriented analysis of the process leading to continued usage intention, the work of Ding et al. [19] contributes technical innovation, focusing on the development and optimization of an integrated mental stress detection system. Conversely, the systematic review by Triantafyllidis et al. [20] synthesizes data from various interventions, highlighting both positive outcomes and practical challenges associated with smartwatch use in real-world clinical settings. The combined approach emerging from these studies emphasizes the necessity for interdisciplinary research, where the integration of theoretical models, technical optimization, and the critical evaluation of the existing literature is essential for the effective exploitation of smartwatch capabilities in digital health. This integrative perspective suggests that psychological processes, technological innovations, and practical applications must all be taken into account in order to achieve a comprehensive solution that improves not only health monitoring but also users’ quality of life [12,19,20].
According to the study by Can, Arnrich, and Ersoy [21], their research provides an extensive overview of efforts to detect stress in daily life through the use of smartphones and wearable devices. Their approach focuses on a range of physiological measurements—such as heart activity monitoring (with emphasis on HRV changes), electrodermal activity (EDA), and other complementary data—that are combined with machine learning techniques to distinguish between stress and relaxation states. The research methodology includes an analysis of different contexts in which the measurements are taken, including laboratory settings, workplaces, car commutes, and everyday unrestricted environments. The authors also thoroughly examine challenges in data collection—such as synchronizing multiple sensors, the limitations of non-invasive measurement methods, and energy conservation issues in real-world applications. Through active analysis and the integration of theoretical and practical dimensions, their study demonstrates the evolving potential of stress detection technologies and paves the way for future interdisciplinary applications in digital health. Additionally, the app includes guided meditation tools that help users lower anxiety through breathing and relaxation techniques. Users can set session duration and breathing cycles, while the app also provides feedback on cardiovascular health (e.g., measuring heart rate before and after sessions). The study’s results showed a decrease in anxiety and immediate improvements in users’ sense of well-being following the meditation sessions.
Moreover, Siirtola [22] explored stress detection using commercially available smartwatch sensors, examining whether anxiety can be detected without the need for specialized sensors like electrodermal activity (EDA). The study found that anxiety can be accurately detected using only built-in smartwatch sensors, such as heart rate monitors and skin temperature sensors. The findings revealed that stress detection can be achieved effectively even without EDA, provided that the right sensor data (e.g., heart rate, skin temperature) and appropriately configured analysis windows are used.
The study of Foord et al. [23] examined the utility of smartwatches in predicting relapses in patients with severe mental illness (SMI). Participants used smartwatches to monitor heart rate, activity, and sleep patterns, with results confirming that EDA and activity data were particularly valuable for tracking changes and enabling early warnings of relapse. Moreover, Pinge et al. [24] conducted a systematic review on wearable devices for anxiety monitoring. Their research examined the full process from data collection and preprocessing to machine learning models for detecting anxiety using sensor data (e.g., ECG, EDA, skin temperature). The study emphasized the need for accurate data collection and highlighted that combining wearable devices with ML techniques yields high-accuracy stress detection models while also calling for a deeper understanding of physiological–psychological correlations and algorithm refinement.
Also, Sheikh et al. [25] explored the use of both wearable and environmental sensors in mental health monitoring, emphasizing how real-time physiological tracking (e.g., cardiorespiratory frequency, skin temperature) can empower patients and enable early diagnosis and adaptive treatment. They stressed the transformative potential of wearables in everyday life for those affected by mental disorders. At the same time, Horde et al. [26] evaluated the performance of smartwatches like Apple Watch Series 6 and Fitbit Sense in stress detection. Through experimental tasks using the Multi-Attribute Task Battery II (MATB-II), data collected from HR and HRV sensors were compared with subjective stress reports using the NASA-TLX index. The results showed that while participants reported significantly higher stress during high workload conditions, smartwatch data (HR and HRV) did not reflect these changes accurately. This suggests that smartwatch cardiac data may not reliably mirror perceived stress, though future enhancements in signal interpretation could improve accuracy.
Additionally, Sharma et al. [27] analyzed the contribution of wearables and IoT in mental well-being. Using activity and sleep data from the DISCover project, they built a logistic regression model that correlated changes in PHQ-9 scores with step count patterns, noting that adding emotional parameters could improve mental health detection accuracy (F1 score = 0.66). Also, Houzangbe et al. and Housangbe [28] studied the integration of wearables in Virtual Reality (VR) environments. Using smart wristbands to provide biofeedback based on heart rate, they found that physiological integration improved the VR user’s experience and engagement, suggesting potential for emotional regulation and stress management.
Ciabattoni et al. [8] developed a real-time stress detection system using smartwatches that monitored physiological signals such as galvanic skin response (GSR), heart rate variability (HRV), and body temperature (BT). Their experimental protocol confirmed that smartwatch-derived data can reliably support the diagnosis and tracking of psychological stress. Also, Kumar et al. [10] introduced Resp-BoostNet, an enhanced neural network model that integrates respiration rate prediction to improve smartwatch-based stress detection accuracy. Using heart rate and GSR data, the model achieved a 94% accuracy rate, outperforming previous techniques.
Also, Cola and Vecchio [29] examined the use of wearables in supporting mental health and well-being by tracking biological data such as sleep quality, physical activity, and alcohol consumption using ECG and PPG sensors. Their results showed that smartwatches are effective tools for the continuous monitoring of mental states, highlighting the value of real-time data for personalized care and stress management.
Yen et al. [30] conducted a randomized controlled trial examining the psychological benefits of wearable technology. The study found that smartwatches significantly improved physical activity and stress management, boosting self-esteem and quality of life more than mobile apps alone. Moreover, Szakonyi et al. [31] explored HRV-based stress detection during Trier Social Stress and Stroop tests using ECG sensors. Their algorithms achieved 96.31% accuracy, suggesting that smartwatches are viable for detecting acute stress in daily life. Also, Nelson et al. [32] compared the HR accuracy of commercial wearables (Apple Watch and Fitbit) to medical ECGs. The results showed acceptable deviations (1.3 bpm for Apple Watch and 9.3 bpm for Fitbit), confirming their utility for non-clinical HR tracking, though limitations remain for medical use.
Jerath et al. [33] investigated HRV monitoring via smartwatches for anxiety management. Their findings support the use of real-time biofeedback apps (e.g., breathing or meditation guidance) tailored to HRV patterns, though sensor accuracy remains an area for future enhancement. At the same time, Fauzi and Yang [34] proposed a continuous stress detection framework using smartwatch data and ensemble classifiers. Using the WESAD dataset [34] and methods like Discrete Wavelet Transform (DWT), their model classified emotional states (stress, joy, and calm) with promising results, particularly for healthcare workers under chronic stress.
According to a study by Weiss et al. [6], their aim was to investigate the utility of sensors in smartphones and smartwatches for biometric recognition and identification using a variety of Activities of Daily Living (ADLs). The research explored several motion sensors, such as accelerometers and gyroscopes, to detect movement patterns used in user identification and device management. The results demonstrated that data from 18 different ADLs were successfully used for user recognition and authentication, paving the way for continuous and automatic biometric identification using mobile devices. Halder et al. [35] focused on the use of Long Short-Term Memory (LSTM) models to detect stress through physiological signals collected from smartwatch sensors, such as heart rate, electrodermal activity (EDA), and skin temperature. Using the “Stress-Predict Dataset” [6] with data from 35 participants wearing the Empatica E4 smartwatch, the study found that the LSTM model achieved a stress detection accuracy of 91.78%, demonstrating the potential of this technology for real-time stress monitoring and early diagnosis.
Toshnazarov et al. [36] explored the use of commercially available smartwatches (e.g., Samsung Watch, Apple Watch) for detecting stress in daily life by monitoring biometric data such as heart rate and combining it with user-reported contextual data. The study involved two stages: In a lab, 26 participants took part in controlled experiments to record physiological stress responses, while in real-world settings, 18 participants were monitored. The method achieved an F1 score of 0.84 in lab conditions and 0.71 in real-life conditions, indicating promising results for real-world stress detection using consumer wearables. In addition, Zhu et al. [4] investigated the use of electrodermal activity (EDA) from wearable sensors (e.g., smartwatches) for stress detection. The study evaluated different machine learning algorithms using four public datasets [4] containing EDA data collected in real-world contexts. Five ML models were tested to classify stress versus non-stress conditions, with the Support Vector Machine (SVM) achieving the highest accuracy at 92.9%.
Figure 3Comparison of machine learning algorithms employed for stress detection based on smartwatch biosignals (e.g., HRV, sweat cortisol), highlighting model accuracy and data integration methods [36].
[Figure omitted. See PDF]
Katarya and Maan [5] conducted a review of various stress detection methods using smartwatches and machine learning algorithms. Their methodology analyzed physiological signals such as HRV, GSR, skin temperature, and sleep changes. Algorithms like SVM, KNN, and Random Forest were used for classification. The findings indicated that combining physiological signals with SVM provided the most accurate stress detection. Also, Amate et al. [37] examined the use of smartwatch data and social media analysis for mental health diagnostics. Their aim was to develop a methodology to detect users’ emotional states based on social media activity and identify abnormal patterns through smartwatch-collected data. Heart rate, stress levels, and other biometrics were analyzed alongside sentiment analysis to support mental health professionals in decision-making. The collective evidence demonstrates that smartwatch technology offers significant promise for stress detection, health monitoring, and well-being, yet it also reveals critical limitations and trade-offs. On one hand, these wearables can continuously capture key physiological indicators—from heart rate variability to electrodermal activity—enabling the early detection of stress-related anomalies and encouraging users toward healthier habits.
This suggests that tailoring stress detection to each user’s physiology can mitigate variability and improve reliability. However, translating these successes to real-world environments remains an ongoing challenge. Studies that moved from the lab to daily life report a drop in model accuracy (e.g., F1 scores declining from 0.84 in controlled tests to 0.71 in free-living conditions), highlighting how factors like context and motion introduce noise that algorithms must learn to handle. Likewise, what constitutes “ground truth” stress is much harder to pin down outside the lab—self-reported stress levels can be subjective and inconsistent, underscoring the need for better methods of labeling and validating stress in natural settings.
Controlled trials echo these positive trends; a randomized trial reported that using a smartwatch led to better stress management, higher physical activity, and increased self-esteem compared to relying on mobile phone apps alone. Such diverse examples—spanning preventive health, chronic illness management, and acute care—underscore that wearable stress monitors, when well-implemented, can facilitate earlier interventions, empower patients in self-management, and potentially improve clinical outcomes. At the same time, they remind us that reported successes often come from pilot studies or specific contexts, and achieving similar benefits at scale will require addressing the practical challenges documented in the related literature. Across all these investigations, a recurring theme is the necessity of an interdisciplinary and user-centered approach in advancing wearable stress detection. The integration of knowledge from engineering, computer science, clinical psychology, and design is not merely beneficial but essential. As noted in one comparative analysis, progress in this field demands combining theoretical models of human behavior, technical innovation in sensing and analytics, and the rigorous evaluation of outcomes into a cohesive research strategy.
In practice, this means that innovation must be balanced with real-world applicability. Cutting-edge sensors and complex algorithms will have limited impact unless they are matched by equal efforts to make devices practical for everyday users and acceptable in healthcare workflows. Many authors point out this balance: For instance, even as new biochemical sensors (such as wrist-mounted cortisol detectors) and flexible electronics are being developed to capture stress markers with unprecedented fidelity, there remains a pressing need to ensure that basic issues like device autonomy, data reliability, and interoperability are solved for use outside the lab. Large-scale trials and longitudinal studies are frequently cited as the next step for verifying that smartwatch-based interventions can maintain their efficacy over time and across diverse populations.
Future research directions therefore center on bridging the gap between prototype and practice. On the technological side, priorities include improving sensor precision (e.g., refining HRV and EDA measurements and incorporating novel biomarkers), reducing power consumption to extend battery life, and enhancing machine learning models through training on more extensive, representative datasets—all to ensure that stress detection remains accurate in the messy complexities of daily life.
Equally crucial is the human factor side: advancing data privacy protections and ethical guidelines, simplifying user interfaces, and tailoring feedback to be actionable without causing alarm or fatigue. The next generation of digital health solutions will likely see smartwatches more deeply integrated with clinical decision systems and therapeutic platforms, as some early systems have begun to carry out by linking wearable data with behavioral coaching or telemedicine.
Finally, the path forward lies in harmonizing innovation with usability—marrying cutting-edge biosensors and AI-driven analytics with careful attention to patient experience, engagement, and safety. By doing so, researchers and practitioners can fully realize the potential of smartwatches as holistic stress management tools, translating the wealth of continuous biometric data into meaningful improvements in mental health and well-being at both individual and population levels.
3.2. Smartwatches for Mental Health Monitoring
The systematic review by Triantafyllidis et al. [20] offers a broad overview of smartwatch-based interventions aimed at improving health and well-being. Following rigorous PRISMA standards, their study analyzes the characteristics of research that incorporates such interventions, focusing on both study design and outcomes. The review shows that most interventions address specific clinical conditions—such as heart disease, diabetes, depression, and stress—and that the use of smartwatches as tools for self-management and vital sign monitoring improves both objective and subjective outcomes for patients. Moreover, the review highlights challenges related to technology, such as the need for frequent charging, the reliability of data collection, and the integration of smartwatches with other devices, all of which affect both the acceptance and practical application of these tools in clinical environments [20].
The study by Nadal et al. [38] also presents an innovative approach by integrating a smartwatch into a web-based intervention for depression. Their research aims to reduce the burden of self-reporting by automating the recording of parameters such as sleep and physical activity, thereby providing high-accuracy real-time data. The methodology includes a structured testing protocol that evaluates a patient’s acceptance of the technology and its impact on therapeutic progress, underscoring the value of non-invasive self-monitoring systems in digital mental health interventions.
Finally, the systematic review by Ahmed et al. [39] explores the potential of wearable devices for monitoring and managing mental health disorders such as depression and anxiety. Drawing on a wide range of studies, the authors highlight the characteristics of available technologies, methodological approaches, and major challenges involved in applying wearable devices to mental healthcare. This analysis notes that although consumer devices are increasingly accessible and embedded in daily life, issues remain concerning the reliability of data collection, scalability, and the accurate measurement of psychological parameters [39].
A consistent picture emerges regarding how wearable devices can contribute to the monitoring and management of stress and mental health. While Can et al. [21] focus on the technical aspects of non-invasive physiological signal detection, the review by Reeder and David [13] maps the applications of smartwatches in health, with emphasis on practical challenges and opportunities for improvement. Nadal et al. [38] expand this perspective by integrating wearable devices into therapeutic interventions for depression, and the systematic review by Ahmed et al. [39] provides a comprehensive evaluation of technological solutions for monitoring mental disorders. This holistic approach highlights the necessity for continuous interdisciplinary collaboration to fully leverage the potential of wearables in mental health, thereby enhancing future therapeutic interventions and digital health applications.
According to Dibia [40], the Foqus app was developed as an innovative solution for managing mental conditions such as Attention Deficit Hyperactivity Disorder (ADHD) using a smartwatch. The main goal of the app is to help individuals improve their focus and reduce anxiety by offering tools based on scientific management strategies. Foqus includes three key features: the Pomodoro technique to enhance focus, guided meditation to reduce stress, and positive messaging to boost mood.
The study also highlighted the importance of window size in data analysis. Longer windows (e.g., 120 s) provided higher accuracy in recognizing emotional states than shorter ones (e.g., 15 s). Specifically, by using Linear Discriminant Analysis (LDA) and combining data from skin temperature, BVP, and ECG sensors, the study achieved an anxiety recognition accuracy of 87.4%.
Similarly, Long et al. [41] evaluated the increasing use of wearables in mental health monitoring for the prediction and early diagnosis of mental disorders. The research assessed the effectiveness of wearable sensors in detecting mental conditions such as anxiety and depression and in analyzing collected data to manage these states. The study concluded that wearables using sensors like ECG and skin temperature can provide accurate readings for detecting and monitoring mental health conditions, especially when supported by advanced AI algorithms.
At the same year, Samyoun et al. [42] introduced M3Sense, a system for representation learning to detect emotional states using wearable sensors. The study aimed to improve mental state detection by using physiological data from wearables. A major challenge addressed was the potential inaccuracy of sensor data due to limited datasets and variation in sensor devices. M3Sense applies Domain Alignment techniques to transfer learned representations across tasks without needing additional data or labels. The results showed that M3Sense outperformed traditional machine learning methods with handcrafted features, improving performance by 9.6% to 26.2% in areas of anxiety and emotion detection.
Also, Hassantabar et al. [7] presented MHDeep, a Deep Neural Network system for the early diagnosis and monitoring of psychiatric disorders such as schizoaffective disorder, major depressive disorder, and bipolar disorder using medical-grade wearable sensors. The models achieved diagnostic accuracy rates of 90.4%, 87.3%, and 82.4%, respectively. Synthetic data generation techniques helped enhance training quality, overcoming the limitations of data availability for these conditions.
Moreover, Nadal et al. [43] focused on the acceptance of smartwatch-based self-monitoring in clinical settings for depression. Their study analyzed user experiences with the Mood Monitor app during an 8-week digital cognitive behavioral therapy (iCBT) intervention with 69 participants. The results showed broad acceptance, with patients reporting that smartwatch-based mood tracking offered more privacy and was less burdensome than handwritten logs, improving engagement in therapy.
In addition, Kaplan et al. [44] studied smartwatch use in promoting healthy behavior in older adults. In a 4-week study with 20 healthy seniors, participants used the Garmin Vívosmart 4. During the final week, they received feedback on their sleep and activity levels. The survey results revealed behavioral improvements and, more specifically, 50% improved their diet, 45% increased physical activity, and 30% reported better sleep. While these results support the potential of wearables in encouraging healthy habits among the elderly, further research is needed to determine whether feedback leads to long-term behavioral change. Also, Long [15] reviewed the broader role of smartwatches in healthcare, especially for physical fitness and mental health in conditions like dementia, depression, and high stress. The study highlighted how smartwatches measure biometric data such as heart rate, blood pressure, and sleep quality and concluded that they are highly useful in supporting individuals with various mental health conditions and improving overall well-being.
Similarly, Rana et al. [15] analyzed smartwatch use across diverse populations—including seniors, individuals with dementia, people with depression, and athletes—and concluded that smartwatches offer valuable tools for daily health management by continuously tracking physical activity, fitness, and mental health.
According to the study by Johnston et al. [23], the integration of smartwatches into mental health care holds significant potential for health monitoring and relapse prediction in individuals with severe mental illnesses. This research focused on four patient cases involved in a study exploring the implementation of mHealth technology for monitoring mental health. The goal was to assess whether biometric data from smartwatches could predict relapses and support clinical decision-making.
The methodology involved the use of the mHealth device Empatica Embrace2, a smartwatch equipped with sensors tracking activity, sleep, cardiorespiratory function, and electrodermal activity (EDA). Patients wore the smartwatches for six months, with real-time data monitored by clinicians. The results showed that changes in sleep and EDA preceded hospitalizations in two cases, indicating that smartwatch data could serve as early warning signs for clinical intervention. In addition, Sara et al. [45] also investigated anxiety detection using smartwatches through HRV (heart rate variability) and EDA. Their study found that smartwatch sensors capturing fluctuations in cardiorespiratory frequency and electrodermal activity provide useful insights for daily anxiety management. However, the authors noted that these devices lack the precision of medical-grade tools.
Shaharudin et al. [23] focused on mental health management via the smartwatch tracking of EDA and cardiorespiratory frequency. While EDA data may not always match subjective stress perception, it still offers valuable indicators when used alongside psychiatric assessments. Moreover, Robinson et al. [46] conducted a systematic review evaluating the role of wearables in the self-management of subclinical mental disorders (e.g., anxiety, depression, sleep disorders). Using biometric data from 12 devices, including EDA and HR, and ML models such as SVM, they found that while wearables provide useful monitoring data, limitations persist in reliably predicting subclinical conditions.
Finally, Lui et al. [47] evaluated the Apple Watch’s ability to monitor physiological indicators related to mental health, such as heart rate and HRV. Figure 4 provides a broader overview of the applications of wearable devices in healthcare, highlighting key biosignals (e.g., heart rate, blood pressure, glucose) and the sensor technologies used to capture them. Although the results showed promise for detecting mild anxiety, the study also highlighted limitations in accuracy during intense movement, calling for further sensor validation and research into Apple Watch applications in mental health.
According to the study by Hassan et al. [48], consumer wearable devices have shown potential for monitoring and predicting mental and physical health in individuals with severe mental illnesses (SMIs). The research focused on how devices such as smartphones and smartwatches can detect indicators associated with psychiatric relapse, such as sleep disturbances, cardiovascular changes, and physical activity patterns. The analysis revealed that although smartwatches can track real-time data, challenges remain in accurately correlating this information with mental health status due to variability in measurement methods and devices. The study concluded that wearables could offer timely and valuable insights for mental health monitoring but require improved understanding and validation for psychiatric use.
At the same time, Huaroto et al. [49] proposed combining cognitive behavioral therapy (CBT) with physiological monitoring through smartwatches for managing anxiety and cardiovascular depression. Their app integrated metrics such as heart rate and blood oxygen saturation, and the results showed over 80% improvement in patient-reported outcomes, with therapists positively evaluating its usefulness for promoting mental well-being. Moreover, Ueafuea et al. [50] examined the impact of the COVID-19 pandemic on mental health, highlighting the increase in anxiety, depression, and insomnia due to social distancing and uncertainty. The authors stressed the importance of digital technologies and wearables in remote mental health support, noting that smartphones and smartwatches can effectively detect and monitor psychological conditions by anonymously tracking indicators like respiration rate, activity, and electrodermal activity (EDA).
All of these studies support the idea that, when used appropriately, smartwatches and wearable devices can provide reliable data for the detection and monitoring of conditions such as anxiety, depression, and insomnia. The integration of this data with machine learning models can improve prediction and allow for early diagnosis and intervention [8,10,50].
Jeong et al. [51] focused on psychological stress during radiotherapy (RT) and how this can be captured using wearables. Patients wore the Galaxy Watch Active 2 during and before sessions to track heart rate and anxiety scores. The results showed increased HR and perceived stress during treatment, especially in initial sessions, indicating that smartwatches can monitor patient mental states and support stress-reduction strategies during therapy. Moreover, Li et al. [10] used cardiopulmonary and GSR data collected via smartwatches to detect stress in real time with the BoostNet algorithm. The model achieved 94% accuracy, demonstrating smartwatches’ capability to provide reliable indicators of mental health in high-stress conditions.
Hickey et al. [52] conducted a review of wearable tech for detecting stress, depression, and anxiety. HRV and EDA were key parameters, but the highest accuracy for anxiety detection was achieved when combined with EEG. EEG-based depression detection remains a challenge due to commercial unavailability. In addition, Chalmers et al. [53] used Fitbit HRV data to predict stress responses in medical students and general population participants during a Trier Social Stress Test (TSST). While patterns varied, HR and HRV significantly changed under stress, supporting HRV as a stress indicator. Moreover, Kang and Chai [54] reviewed biochemical markers and sensor technology for wearable-based mental health monitoring. Despite technological progress, they emphasized the need for improved accuracy and reliability in detecting stress-related markers.
According to the study of Machado-Jaimes et al. [55], a well-being monitoring system using physiological and psychological data from smartwatches and questionnaires was developed. Among the six ML algorithms that had been tested, Random Forest achieved the best performance (88% accuracy) in predicting user well-being. Also, Gomes et al. [56] evaluated wearable sensors for the real-time detection of anxiety and panic. Using ML algorithms on HR, EDA, and skin temperature data, the study emphasized wearables’ potential in personalized stress intervention, though accuracy improvements are needed.
Finally, the important study of Dai et al. [9] focused on flexible wearable electronics for continuous health monitoring, particularly for early diagnosis and disease prevention. The study assessed advances in flexible sensors for tracking body temperature, heart rate, and blood pressure. It included analysis of detection mechanisms (e.g., piezoelectric, thermoelectric) and material design. The findings highlighted their potential for continuous home-based medical applications. At the same time, Jiang et al. [11] developed a wearable device for monitoring mental well-being using both social and physiological signals. The study aimed to understand the relationship between physical and mental states by collecting data on behavior, physiology, and audio. The results showed that the device could offer valuable insights into mental health monitoring, particularly for autism and anxiety contexts.
Indeed, participants in digital mental health programs have reported that smartwatch-based mood and stress tracking feels less intrusive than traditional methods—offering greater privacy and convenience—which in turn improves their engagement with therapy.
Methodologically, researchers have begun addressing such issues through innovative means: synchronizing data from multiple sensors and contexts, experimenting with data processing choices (for instance, using longer monitoring windows of around 2 min, which significantly boosts detection accuracy compared to 15 s snapshots), and even generating synthetic data to augment limited training datasets. These efforts reflect a maturing intersection of biomedical engineering and data science, where improving signal quality and algorithmic robustness go hand in hand. Still, certain limitations are sobering—for example, one study found that popular smartwatches’ heart rate and HRV data did not reliably reflect users’ reported stress during complex multitasking scenarios, a caution that even core biometric signals may fail to capture the nuances of psychological stress. Taken together, these findings advocate for a careful empirical approach to algorithm development and validation, ensuring that stress-detection models are trained on diverse, ecologically valid data and tested against meaningful benchmarks of mental well-being. Encouragingly, numerous interventions and clinical studies have demonstrated that when these technological and human factors align, the use of smartwatches can yield tangible health benefits.
In clinical mental health care, for example, continuous monitoring via wearables has shown potential for early intervention. Preliminary case studies of patients with severe mental illness revealed that subtle changes in smartwatch-recorded metrics—such as deteriorations in sleep patterns and spikes in electrodermal activity—preceded clinical relapse events and hospitalizations. These behavioral gains illustrate how real-time feedback from wearable sensors can translate into healthier routines in daily life. Integrating smartwatches with therapeutic interventions also amplifies their impact. One study combining a smartwatch app with cognitive-behavioral therapy for anxiety found over 80% improvement in patient-reported outcomes, with clinicians praising the system’s utility in promoting mental well-being.
In a related vein, patients undergoing digital therapy for depression accepted smartwatch mood monitoring enthusiastically, noting that it felt more private and less burdensome than manual mood diaries—a convenience that improved their engagement and adherence to treatment. Even in high-stress medical situations, wearables prove beneficial: for instance, cancer patients wearing a smartwatch during radiotherapy sessions showed elevated heart rates correlating with anxiety spikes, indicating these devices can effectively monitor patient stress and guide timely stress-reduction strategies during care.
3.3. Machine Learning Algorithms for Stress and Mental Health Management in Smartwatches
This section provides a short summary of machine learning (ML) algorithms used in smartwatches for mental health monitoring and stress management. Specifically, smartwatch-based systems for mental health monitoring utilize a variety of machine learning algorithms designed to interpret physiological and behavioral signals. Common choices such as k-Nearest Neighbors (KNNs), Decision Trees (DTs), and Random Forests (RFs) offer interpretable and efficient classification, especially in structured biometric data. Support Vector Machines (SVMs), particularly with Gaussian kernels, are widely applied due to their accuracy in handling complex and nonlinear biosignals.
Naive Bayes and Logistic Regression provide fast, lightweight models that are often used in early-stage or real-time detection tasks. In contrast, Artificial Neural Networks (ANNs) and Deep Neural Networks (DNNs) are suited for capturing deeper, nonlinear relationships in multi-modal sensor data. Boosting Neural Networks (BNNs) combine multiple weak learners to enhance predictive performance.
Dimensionality reduction and feature selection techniques such as Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and Bayesian Networks (BNs) further support the modeling process. In long-term social sensing, heuristic methods like signal magnitude area (SMA) and short-term audio energy thresholding are also employed to infer behavior patterns relevant to mental health.
Collectively, these algorithms support the accurate real-time inference of stress and mental states, leveraging data from wearable devices in both clinical and everyday settings.
As shown in Figure 5, the reported precision values generally range from 70 to 85%, with AUROC values only occasionally provided. This lack of standardized reporting further highlights the challenge of quantitative comparison between studies.
3.4. Algorithm Performance Comparison
To improve readability and facilitate direct comparison between studies, Table 2 summarizes the machine learning models employed in the reviewed literature. For each study, we report the type of algorithm, biometric features used, ground truth strategy, and performance metrics such as AUROC or accuracy. This structured overview supports the interpretation of methodological trends and highlights variability in predictive performance.
4. Discussion
4.1. Interpretation of Key Findings
The results of this systematic review highlight three main insights. First, smartwatches demonstrate promising capabilities in detecting stress and monitoring mental health through physiological signals such as heart rate variability, electrodermal activity, and motion data. In studies, several machine learning algorithms, including SVM, Random Forest, and LSTM, have achieved moderate to high accuracy (AUROC > 0.75) in predicting stress levels, validating the feasibility of real-time emotion detection with wearable devices.
Second, the practical adoption of smartwatches for health monitoring depends not only on technical accuracy but also on user perceptions of comfort, intrusiveness, and usefulness. As shown in studies such as Papa et al., the perceived ease of use and device ergonomics significantly influence adoption intentions. These findings align with established technology acceptance models and suggest that even high-performance algorithms may have limited impact if devices are not perceived as wearable or trustworthy.
Third, our findings suggest that personalization improves performance in stress detection. Studies that tailored algorithms to individual baseline physiology showed higher predictive power compared to generic models. This reinforces the need for user-specific adaptive systems in mental health applications rather than a one-size-fits-all approach.
In general, these findings underscore the dual importance of algorithmic robustness and user-centered design in achieving the full potential of smartwatch-based mental health tools.
4.2. Comparison with the Previous Literature
Our review confirms and extends key observations from recent individual studies. The findings by Dai et al. [3] demonstrated that commercially available smartwatches, when combined with machine learning, can reliably distinguish stress states in controlled environments. Our synthesis of additional studies supports this result and shows similar predictive performance across various contexts and algorithms, thus reinforcing the generalizability of their conclusions.
Furthermore, our review echoes the conclusions of Dhar et al. [1], who emphasized the value of continuous physiological monitoring for proactive health management. We expand this perspective by highlighting not only health tracking, but also mental health prediction capabilities as a growing field of application for smartwatches.
The work of Papa et al. [2] highlighted how perceived comfort and device intrusiveness affect user attitudes toward the adoption of smartwatches. Our findings support their model and show that these perceptions indirectly shape acceptance by influencing perceived usefulness and ease of use. However, our review adds nuance by showing that, even with ergonomic devices, real-world adoption may be hindered by ethical and privacy concerns.
Overall, our review integrates and expands upon prior studies by connecting algorithmic performance with user acceptance and contextual constraints, offering a broader understanding of how smartwatch-based systems perform both technically and practically.
4.3. Practical Implications
The findings of this review suggest that smartwatches can serve as effective tools for early stress detection and mental health support in everyday settings. This has meaningful implications for preventive healthcare, particularly in environments where access to mental health professionals is limited. Smartwatches offer a low-cost, non-invasive means to continuously monitor psychological well-being and could be integrated into digital health platforms to alert users—or clinicians—when early signs of stress emerge.
From a design perspective, the review emphasizes the importance of personal comfort, perceived usefulness, and trust in data privacy. Developers and manufacturers must prioritize ergonomic form factors, intuitive interfaces, and transparent data handling policies to maximize adoption. Furthermore, machine learning systems for stress detection should allow for personalized calibration based on individual baseline data, which has been shown to improve predictive accuracy.
Healthcare providers and digital health startups can use these insights to design interventions that combine passive monitoring with active guidance, such as delivering personalized coping strategies when stress is detected. Finally, policymakers must address ethical and legal concerns, particularly around data ownership, informed consent, and the medical validity of algorithmic diagnoses, to ensure safe and responsible use of smartwatch-based health technologies.
4.4. Limitations
This systematic review has several limitations that should be acknowledged. First, the studies included in the analysis vary widely in their methodologies, populations, and sensor configurations. This heterogeneity limits the ability to generalize findings or directly compare algorithmic performance across studies.
Second, many studies relied on subjective self-reports to define ground truth for stress or mental health states. While such labels are common, they introduce variability and potential bias, especially when no physiological benchmarks are used alongside them.
Third, the majority of included research was conducted in short-term or laboratory-controlled settings. As a result, the long-term feasibility and reliability of smartwatch-based stress detection in real-world conditions remain underexplored.
Fourth, few studies explicitly addressed issues related to algorithm interpretability or transparency. Most models are treated as black boxes, which could hinder their acceptance in clinical or public health contexts.
Finally, publication bias and the exclusion of non-English or unpublished data may have resulted in missing relevant studies, especially from underrepresented populations or regions.
4.5. Quality Assessment and Meta-Analysis
This review did not include a formal quality assessment using tools such as ROBINS-I or GRADE. These tools require comparable study designs and consistent outcome measures, which were not present across the included studies. The reviewed research varied widely in methodology, sensor setups, validation procedures, and definitions of stress or mental health outcomes. Because of this heterogeneity, applying a unified risk of bias tool would not have produced meaningful results. As the field progresses and research designs become more standardized, the use of formal quality assessment tools will become increasingly relevant.
In addition, despite the fact that a quantitative synthesis could strengthen the overall conclusions of this review, a meta-analysis was not feasible. The primary reason was the significant heterogeneity across the included studies, both in terms of methodological design and reported outcomes. Studies varied in algorithmic models, sensor configurations, data collection durations, and outcome definitions. These inconsistencies prevented the meaningful aggregation of effect sizes or performance metrics.
Instead, this review provides a structured qualitative synthesis, which allows for identifying common trends and challenges in smartwatch-based stress and mental health detection. Future work may consider more focused subgroups or standardized protocols to enable quantitative comparisons across studies.
4.6. Ethical and Clinical Considerations
Smartwatch-based mental health monitoring raises significant ethical and clinical challenges. While these devices support early detection through continuous biometric tracking, they also collect highly sensitive data—such as stress levels, sleep patterns, and heart rate variability—that require careful handling.
A central ethical concern involves data privacy and security. Without strong safeguards, personal health information may be exposed to misuse by insurers, employers, or unauthorized third parties. Developers must implement secure data storage, anonymization procedures, including secure storage, anonymization protocols, and compliance with privacy regulations such as the General Data Protection Regulation (GDPR [60]) and HIPAA.
From a clinical standpoint, smartwatch data offers opportunities for remote assessment and early intervention, particularly in underserved populations. However, these benefits depend on rigorous validation. Without clinical trials or regulatory approval (e.g., CE marking or FDA clearance), such devices cannot replace standardized diagnostic methods or inform medical decisions reliably.
Another concern is how users interpret feedback from these devices. Without clinical guidance, users may misread outputs or rely too heavily on them for self-diagnosis. Developers must ensure that the information presented is clear, contextualized, and supports user autonomy rather than inducing anxiety or false reassurance.
Ultimately, the ethical integration of smartwatch technologies into mental health care depends not only on technical accuracy but also on responsible design, clinical collaboration, and transparency in communication.
5. Conclusions
However, technical constraints persist: Measurement reliability can be skewed by motion artifacts or skin differences, battery life remains limited for uninterrupted monitoring, and the sensitive data collected raise serious privacy concerns. Thus, while smartwatches are valuable adjuncts to health surveillance, their efficacy hinges on improving the underlying technology and ensuring that data are managed securely and ethically. Researchers emphasize that current devices must advance in sensor accuracy and power efficiency before they can be safely and reliably integrated into mainstream healthcare.
This duality of potential and challenge underscores the need for a balanced, rigorous approach in developing smartwatch-based stress interventions. From an ergonomic and psychological standpoint, user acceptance emerges as a deciding factor in the real-world impact of these tools. Technological sophistication alone is not enough—devices must also fit seamlessly into daily life. Empirical studies show that when a smartwatch is comfortable and unobtrusive, users tend to perceive it as more useful and easy to use, indirectly bolstering their intention to adopt the device. In other words, improving the wearability and usability of stress-tracking watches can enhance their perceived value and encourage consistent use. Moreover, the psychological benefits users associate with wearables can reinforce engagement. For instance, if a smartwatch helps boost productivity, provides motivation, or enhances one’s self-image, it can activate positive feelings of inspiration and well-being that feed into continued usage.
These findings highlight that successful stress-monitoring devices must be designed not only with high-tech sensors but also with human factors in mind. Ergonomic design, intuitive interfaces, and clear personal benefits are all critical to fostering user trust and sustained use, without which even the best technology cannot translate into health improvements. Equally important are the technological and methodological considerations that determine how effectively smartwatches can detect stress in practice. A range of machine learning approaches has been applied to wearable sensor data, and the results are often encouraging. In controlled settings, models can distinguish stress from non-stress with notable accuracy: for example, a Support Vector Machine trained on heart-rate patterns achieved an AUROC of 0.79 in detecting stress episodes, and incorporating additional signals or advanced algorithms can push accuracy even higher (often into the 90%+ range). Sophisticated techniques like Deep Neural Networks and representation learning have further improved the detection of mental states, in some cases outperforming traditional classifiers by margins of 10–25%. Notably, the personalization of models shows great promise—calibrating algorithms to an individual’s baseline can raise performance to levels comparable to lab-grade systems.
By flagging these precursors, wearables acted as early warning systems, enabling care teams to intervene sooner. Similarly, another investigation confirmed that trends in EDA and activity data were valuable for tracking worsening conditions, reinforcing that such signals can serve as actionable indicators of relapse risk. Beyond acute psychiatry, wearable devices have been tested in broader populations with notable outcomes. Among older adults, a simple 4-week program of wearing a fitness smartwatch and receiving feedback led to self-reported improvements in lifestyle: roughly half the participants adopted healthier diets, nearly as many increased their physical activity, and a third experienced better sleep.
As a concluding remark, the growing body of interdisciplinary research underscores the significant potential of smartwatches as tools for mental health monitoring and stress management. While current limitations—such as sensor accuracy, data privacy concerns, motion artifacts, battery constraints, and inconsistent usability—pose real challenges to clinical adoption, ongoing advancements in wearable sensing, algorithm design, and user-centered development offer promising avenues for improvement. Beyond technical progress, user acceptance remains essential: Comfortable and intuitive devices are more likely to foster long-term engagement and perceived benefit. Studies already demonstrate meaningful outcomes, from the early detection of psychiatric relapses to healthier lifestyles in older adults. To fully harness the value of these tools, future work must prioritize ergonomic design, robust validation, regulatory alignment, and ethical safeguards. With these foundations in place, smartwatch-based interventions can evolve into credible assets in personalized care and scalable mental health support.
Despite the progress made, several limitations hinder the widespread adoption of smartwatch-based mental health applications. The key challenges include data heterogeneity, lack of standardized validation protocols, and concerns around privacy and data security. To address these, future research should aim to establish shared datasets and evaluation standards, improve interoperability across devices, and ensure the transparent reporting of algorithm performance. Additionally, strong ethical frameworks, user trust, and practical usability must be prioritized to translate research into real-world impact.
Conceptualization, N.-A.K.; methodology, N.-A.K. and A.S.; validation N.-A.K. and A.S.; formal analysis, N.-A.K. and A.S.; investigation, N.-A.K. and A.S.; resources, N.-A.K. and A.S.; data curation, N.-A.K. and A.S.; writing—original draft preparation, N.-A.K. and A.S.; writing—review and editing, C.-N.A. Anagnostopoulos. N.-A.K. and A.S.; supervision, C.-N.A. Anagnostopoulos. All authors have read and agreed to the published version of the manuscript.
No new data were created or analyzed in this study. Data sharing is not applicable to this article.
The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.
Footnotes
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
Figure 1 Sequence of activities for the laboratory-based phase of this study. (A) Timeline of the experiment including stress-inducing tasks (e.g., speech, math, cold pressor test), self-report periods, and resting phases. (B) Fossil Gen4 Explorist smartwatch instrumented for this study. (C) Computer-based math tasks used for this study [
Figure 2 PRISMA 2020 flow diagram illustrating the identification, screening, eligibility assessment, and inclusion process for this systematic review. A total of 856 records were retrieved through systematic searches in PubMed, Scopus, and IEEE Xplore using the following query: (“smartwatch” OR “wearable”) AND (“stress” OR “mental health”). After removing duplicates, 719 records were excluded during title and abstract screening. The remaining 137 full-text articles were assessed against predefined inclusion and exclusion criteria, resulting in the exclusion of 75 articles. Ultimately, 61 studies were included in the qualitative synthesis.
Figure 4 Overview of wearable device applications in healthcare monitoring [
Figure 5 Performance of ML models in smartwatch-based stress/mental health studies, including accuracy and AUROC where available. This figure illustrates the variability of performance metrics across different studies [
Overview of studies, algorithms used, and study purpose.
| Study Title | Algorithms Used | Purpose |
|---|---|---|
| Zhu et al. [ | SVM, Naive Bayes, Logistic Regression, Random Forest, KNN | Stress detection using EDA signals. Algorithms were chosen based on their ability to manage multidimensional and nonlinear features. |
| Katarya et al. [ | SVM, KNN, Random Forest, Logistic Regression | Evaluation of different signals (HRV, GSR, ST) and algorithms for smartwatch-based stress detection. |
| Weiss et al. [ | k-Nearest Neighbors, Decision Tree, Random Forest | Biometric identification through daily activity recognition using mobile/time sensors. |
| Hassantabar et al. [ | Deep Neural Networks | Classification between healthy and disordered individuals (e.g., schizophrenia, MDD, bipolar) using wearable data. |
| Ciabattoni et al. [ | k-Nearest Neighbors, Decision Tree, Random Forest | Motion-based identification models using smartwatch sensor data. |
| Dai et al. (2022) [ | PCA, LDA, SVM, KNN, RF, LR, ANN, BN, DT | Comparative accuracy evaluation of algorithms in mental health diagnosis. |
| Ciabattoni et al. [ | Neural Network | Stress level classification using GSR, RR interval, and temperature. |
| Kumar et al. [ | BNN, ANN | Respiration rate prediction and final stress level classification using Boosting techniques. |
| Jiang et al. [ | SMA, short-term audio energy thresholding | Social sensing through speech and activity signal analysis to detect behavioral patterns associated with mental health. |
Summary of ML model performance in smartwatch-based stress or mental health studies.
| Study | Algorithm | Features Used | Ground Truth | AUROC/Accuracy |
|---|---|---|---|---|
| Dai et al. [ | SVM | HR, HRV, EDA | Objective + Subjective | 0.79/0.72 |
| Rohani et al. [ | Random Forest | HRV, Skin Temp., EDA, Accel. | Expert labels | 82.5% |
| Can et al. [ | LSTM | HR, HRV | Self-report + timestamp | 84.6% |
| Gjoreski et al. [ | Decision Tree | HR, HRV, Skin Conductance | Lab-induced | 78.2% |
| Munoz et al. [ | CNN + SVM | HRV, Resp. Rate, Motion | Emotion annotation | 0.76 |
1. Dhar, R.; Kumar, A.; Karmakar, S. Smart wearable devices for real-time health monitoring. Asian J. Med. Sci.; 2023; 14, pp. 1-3. [DOI: https://dx.doi.org/10.3126/ajms.v14i7.55177]
2. Papa, A.; Mital, M.; Pisano, P.; Del Giudice, M. E-health and wellbeing monitoring using smart healthcare devices: An empirical investigation. Technol. Forecast. Soc. Change; 2020; 153, 119226. [DOI: https://dx.doi.org/10.1016/j.techfore.2018.02.018]
3. Dai, R.; Lu, C.; Yun, L.; Lenze, E.; Avidan, M.; Kannampallil, T. Comparing stress prediction models using smartwatch physiological signals and participant self-reports. Comput. Methods Progr. Biomed.; 2021; 208, 106207. [DOI: https://dx.doi.org/10.1016/j.cmpb.2021.106207] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34161847]
4. Zhu, L.; Spachos, P.; Ng, P.C.; Yu, Y.; Wang, Y.; Plataniotis, K.; Hatzinakos, D. Stress detection through wrist-based electrodermal activity monitoring and machine learning. IEEE J. Biomed. Health Inform.; 2023; 27, pp. 2155-2165. [DOI: https://dx.doi.org/10.1109/JBHI.2023.3239305]
5. Katarya, R.; Maan, S. Stress detection using smartwatches with machine learning: A survey. Proceedings of the 2020 International Conference on Electronics and Sustainable Communication Systems (ICESC); Coimbatore, India, 2–4 July 2020; IEEE: Piscataway, NJ, USA, 2020; pp. 306-310.
6. Weiss, G.M.; Yoneda, K.; Hayajneh, T. Smartphone and smartwatch-based biometrics using activities of daily living. IEEE Access; 2019; 7, pp. 133190-133202. [DOI: https://dx.doi.org/10.1109/ACCESS.2019.2940729]
7. Hassantabar, S.; Zhang, J.; Yin, H.; Jha, N.K. MHDeep: Mental health disorder detection system based on wearable sensors and artificial neural networks. ACM Trans. Embed. Comput. Syst.; 2022; 21, 81. [DOI: https://dx.doi.org/10.1145/3527170]
8. Ciabattoni, L.; Ferracuti, F.; Longhi, S.; Pepa, L.; Romeo, L.; Verdini, F. Real-time mental stress detection based on smartwatch. IEEE Sens. J.; 2021; 21, pp. 7162-7172. [DOI: https://dx.doi.org/10.1109/ICCE.2017.7889247]
9. Dai, B.; Gao, C.; Xie, Y. Flexible wearable devices for intelligent health monitoring. View; 2022; 3, 20220027. [DOI: https://dx.doi.org/10.1002/VIW.20220027]
10. Kumar, S.; Chauhan, A.R.; Kumar, A.; Yang, G. Resp-BoostNet: Mental Stress Detection from Biomarkers Measurable by Smartwatches using Boosting Neural Network Technique. IEEE Access; 2024; 12, pp. 149861-149874. [DOI: https://dx.doi.org/10.1109/ACCESS.2024.3461588]
11. Jiang, L.; Gao, B.; Gu, J.; Chen, Y.; Gao, Z.; Ma, X.; Kendrick, K.M.; Woo, W.L. Wearable long-term social sensing for mental wellbeing. IEEE Sens. J.; 2018; 19, pp. 8532-8542. [DOI: https://dx.doi.org/10.1109/JSEN.2018.2877427]
12. Chuah, S.H.W. You inspire me and make my life better: Investigating a multiple sequential mediation model of smartwatch continuance intention. Telemat. Inform.; 2019; 43, 101245. [DOI: https://dx.doi.org/10.1016/j.tele.2019.101245]
13. Reeder, B.; David, A. Health at hand: A systematic review of smart watch uses for health and wellness. J. Biomed. Inform.; 2016; 63, pp. 269-276. [DOI: https://dx.doi.org/10.1016/j.jbi.2016.09.001]
14. William, W.; Wiryawan, D.; Shaharudin, M.S. Feature Factors That Influence Human Health in Smartwatches. J. Inf. Syst.; 2024; 38, pp. 417-432. [DOI: https://dx.doi.org/10.1109/ISIEA61920.2024.10607243]
15. Long, K.Y.; Shanmugam, K.; Rana, M.E. An evaluation of smartwatch contribution in improving human health. IEEE Access; 2024; 12, pp. 400-410. [DOI: https://dx.doi.org/10.1109/IMCOM56909.2023.10035639]
16. Varghese, J.; Alen, C.M.v.; Fujarski, M.; Schlake, G.S.; Sucker, J.; Warnecke, T.; Thomas, C. Sensor validation and diagnostic potential of smartwatches in movement disorders. Sensors; 2021; 21, 3139. [DOI: https://dx.doi.org/10.3390/s21093139] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33946494]
17. Majumder, S.; Deen, M.J. Smartphone sensors for health monitoring and diagnosis. Sensors; 2019; 19, 2164. [DOI: https://dx.doi.org/10.3390/s19092164]
18. Torrado, J.C.; Gomez, J.; Montoro, G. Emotional self-regulation of individuals with autism spectrum disorders: Smartwatches for monitoring and interaction. Sensors; 2017; 17, 1359. [DOI: https://dx.doi.org/10.3390/s17061359]
19. Ding, Y.; Tan, K.; Sheng, L.; Ren, H.; Su, Z.; Yang, H.; Zhang, X.; Li, J.; Hu, P. Integrated mental stress smartwatch based on sweat cortisol and HRV sensors. Biosens. Bioelectron.; 2024; 265, 116691. [DOI: https://dx.doi.org/10.1016/j.bios.2024.116691]
20. Triantafyllidis, A.; Kondylakis, H.; Katehakis, D.; Kouroubali, A.; Alexiadis, A.; Segkouli, S.; Votis, K.; Tzovaras, D. Smartwatch interventions in healthcare: A systematic review of the literature. Int. J. Med. Inform.; 2024; 190, 105560. [DOI: https://dx.doi.org/10.1016/j.ijmedinf.2024.105560]
21. Can, Y.S.; Arnrich, B.; Ersoy, C. Stress detection in daily life scenarios using smart phones and wearable sensors: A survey. J. Biomed. Inform.; 2019; 92, 103139. [DOI: https://dx.doi.org/10.1016/j.jbi.2019.103139]
22. Siirtola, P. Continuous stress detection using the sensors of commercial smartwatch. Proceedings of the 2019 ACM International Joint Conference on Pervasive and Ubiquitous Computing and the 2019 International Symposium on Wearable Computers; London, UK, 9–13 September 2019; pp. 1197-1200. [DOI: https://dx.doi.org/10.1145/3341162.3344831]
23. Foord, R.; Johnston, D.; Casties, A.; Viaphay, N.; Tohamy, A.; Van Leeuwen, N.; Sinclair, J.; Talbot, D.; Harris, A. Integrating smartwatches in community mental health services for severe mental illness for detecting relapse and informing future intervention: A case series. Early Interv. Psychiatry; 2024; 18, pp. 471-477. [DOI: https://dx.doi.org/10.1111/eip.13529]
24. Pinge, A.; Gad, V.; Jaisighani, D.; Ghosh, S.; Sen, S. Detection and monitoring of stress using wearables: A systematic review. Front. Comput. Sci.; 2024; 6, 1478851. [DOI: https://dx.doi.org/10.3389/fcomp.2024.1478851]
25. Sheikh, M.; Qassem, M.; Kyriacou, P.A. Wearable, environmental, and smartphone-based passive sensing for mental health monitoring. Front. Digit. Health; 2021; 3, 662811. [DOI: https://dx.doi.org/10.3389/fdgth.2021.662811]
26. Horde, L.; Moore, A.; Riggs, S.L. Is it Stress O’Clock?: Comparing Cardiac Data from Industry-Leading Smartwatches against Subjective Workload Measures. Proceedings of the Human Factors and Ergonomics Society Annual Meeting; Los Angeles, CA, USA, 13–17 October 2023; Volume 67, pp. 2293-2299.
27. Sharma, P.; Jyoti, J. Role of IoT and data analysis in determining mental well-being. Int. J. Health Sci.; 2022; III, pp. 8140-8144. [DOI: https://dx.doi.org/10.53730/ijhs.v6nS3.7945]
28. Houzangbe, S.; Christmann, O.; Gorisse, G.; Richir, S. Integrability and reliability of smart wearables in virtual reality experiences: A subjective review. Proceedings of the Virtual Reality International Conference–Laval; Virtual, 4–6 April 2018; pp. 1-6.
29. Vecchio, A.; Mulas, F.; Cola, G. Posture Recognition Using the Interdistances Between Wearable Devices. IEEE Sens. Lett.; 2017; 1, pp. 1-4. [DOI: https://dx.doi.org/10.1109/LSENS.2017.2726759]
30. Yen, H.Y. Smart wearable devices as a psychological intervention for healthy lifestyle and quality of life: A randomized controlled trial. Qual. Life Res.; 2021; 30, pp. 791-802. [DOI: https://dx.doi.org/10.1007/s11136-020-02680-6]
31. Szakonyi, B.; Vassányi, I.; Schumacher, E.; Kósa, I. Efficient methods for acute stress detection using heart rate variability data from Ambient Assisted Living sensors. Biomed. Eng. Online; 2021; 20, pp. 1-19. [DOI: https://dx.doi.org/10.1186/s12938-021-00911-6]
32. Nelson, B.W.; Low, C.A.; Jacobson, N.; Areán, P.; Torous, J.; Allen, N.B. Guidelines for wrist-worn consumer wearable assessment of heart rate in biobehavioral research. NPJ Digit. Med.; 2020; 3, 90. [DOI: https://dx.doi.org/10.1038/s41746-020-0297-4]
33. Jerath, R.; Syam, M.; Ahmed, S. The future of stress management: Integration of smartwatches and HRV technology. Sensors; 2023; 23, 7314. [DOI: https://dx.doi.org/10.3390/s23177314]
34. Fauzi, M.A.; Yang, B. Continuous stress detection of hospital staff using smartwatch sensors and classifier ensemble. pHealth 2021; IOS Press: Amsterdam, The Netherlands, 2021; pp. 245-250.
35. Halder, N.; Setu, J.H.; Rafid, L.; Islam, A.; Amin, M.A. Smartwatch-Based Human Stress Diagnosis Utilizing Physiological Signals and LSTM-Driven Machine Intelligence. Proceedings of the 2024 Advances in Science and Engineering Technology International Conferences (ASET); Abu Dhabi, United Arab Emirates, 3–5 June 2024; IEEE: Piscataway, NJ, USA, 2024; pp. 1-8.
36. Toshnazarov, K.; Lee, U.; Kim, B.H.; Mishra, V.; Najarro, L.A.C.; Noh, Y. SOSW: Stress Sensing with Off-the-shelf Smartwatches in the Wild. IEEE Internet Things J.; 2024; 11, pp. 21527-21545. [DOI: https://dx.doi.org/10.1109/JIOT.2024.3375299]
37. Amate, N.; Patil, S.; Jojan, P.; Morankar, S. Use of social media and smartwatch data analytics for mental health diagnosis. Proceedings of the 2021 International Conference on Innovative Trends in Information Technology (ICITIIT); Kottayam, India, 11–12 February 2021; IEEE: Piscataway, NJ, USA, 2021; pp. 1-6.
38. Nadal, C.; Earley, C.; Enrique, A.; Vigano, N.; Sas, C.; Richards, D.; Doherty, G. Integration of a smartwatch within an internet-delivered intervention for depression: Protocol for a feasibility randomized controlled trial on acceptance. Contemp. Clin. Trials; 2021; 103, 106323. [DOI: https://dx.doi.org/10.1016/j.cct.2021.106323] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33621632]
39. Ahmed, A.; Aziz, S.; Alzubaidi, M.; Schneider, J.; Irshaidat, S.; Serhan, H.A.; Abd-Alrazaq, A.A.; Solaiman, B.; Househ, M. Wearable devices for anxiety & depression: A scoping review. Comput. Methods Progr. Biomed. Update; 2023; 3, 100095.
40. Dibia, V. FOQUS: A smartwatch application for individuals with ADHD and mental health challenges. Proceedings of the 2016 ACM International Conference on Interactive Surfaces and Spaces; Niagara Falls, ON, Canada, 6–9 November 2016; pp. 309-314. [DOI: https://dx.doi.org/10.1145/2982142.2982207]
41. Long, N.; Lei, Y.; Peng, L.; Xu, P.; Mao, P. A scoping review on monitoring mental health using smart wearable devices. Math. Biosci. Eng.; 2022; 19, pp. 7899-7919. [DOI: https://dx.doi.org/10.3934/mbe.2022369]
42. Samyoun, S.; Islam, M.M.; Iqbal, T.; Stankovic, J. M3Sense: Affect-agnostic multitask representation learning using multimodal wearable sensors. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol.; 2022; 6, 73. [DOI: https://dx.doi.org/10.1145/3534600]
43. Nadal, C.; Earley, C.; Enrique, A.; Sas, C.; Richards, D.; Doherty, G. Patient acceptance of self-monitoring on a smartwatch in a routine digital therapy: A mixed-methods study. ACM Trans. Comput. Hum. Interact.; 2023; 31, 3. [DOI: https://dx.doi.org/10.1145/3617361]
44. Kaplan, M.; McKniff, M.; Holmqvist, S.; Simone, S.; Rosahl, M.; Chaturvedi, R.; Giovannetti, T. Influence of smartwatch feedback on perceived health behavior change in older adults. Arch. Clin. Neuropsychol.; 2023; 38, pp. 1165-1174. [DOI: https://dx.doi.org/10.1093/arclin/acad067.022]
45. Sara, M.; Bhoomika, V.; Desai, R.V.; Shreyas, S.; Kumaraswamy, S. Stress detection smartwatch. Int. J. Res. Appl. Sci. Eng. Technol.; 2022; 10, pp. 3796-3801. [DOI: https://dx.doi.org/10.22214/ijraset.2022.45865]
46. Robinson, T.; Condell, J.; Ramsey, E.; Leavey, G. Self-management of subclinical common mental health disorders (anxiety, depression and sleep disorders) using wearable devices. Int. J. Environ. Res. Public Health; 2023; 20, 2636. [DOI: https://dx.doi.org/10.3390/ijerph20032636]
47. Lui, G.Y.; Loughnane, D.; Polley, C.; Jayarathna, T.; Breen, P.P. The Apple Watch for monitoring mental health–related physiological symptoms: Literature review. JMIR Ment. Health; 2022; 9, e37354. [DOI: https://dx.doi.org/10.2196/37354]
48. Hassan, L.; Milton, A.; Sawyer, C.; Casson, A.J.; Torous, J.; Davies, A.; Ruiz-Yu, B.; Firth, J. Utility of Consumer-Grade Wearable Devices for Inferring Physical and Mental Health Outcomes in Severe Mental Illness: Systematic Review. JMIR Ment. Health; 2025; 12, e65143. [DOI: https://dx.doi.org/10.2196/65143]
49. Huaroto, L.; Wong, L.; Alvarado, V. Mobile application: For anxiety and cardiovascular depression monitoring using a smartwatch based on cognitive behavioral therapy. Proceedings of the 2022 32nd Conference of Open Innovations Association (FRUCT); Tampere, Finland, 9–11 November 2022; IEEE: Piscataway, NJ, USA, 2022; pp. 112-120.
50. Ueafuea, K.; Boonnag, C.; Sudhawiyangkul, T.; Leelaarporn, P.; Gulistan, A.; Chen, W.; Mukhopadhyay, S.C.; Wilaiprasitporn, T.; Piyayotai, S. Potential applications of mobile and wearable devices for psychological support during the COVID-19 pandemic: A review. IEEE Sens. J.; 2020; 21, pp. 7162-7178. [DOI: https://dx.doi.org/10.1109/JSEN.2020.3046259]
51. Jeong, S.; Jeon, C.; Lee, D.; Park, W.; Pyo, H.; Han, Y. Evaluating psychological anxiety in patients receiving radiation therapy using smartwatch. Radiat. Oncol. J.; 2024; 42, 148. [DOI: https://dx.doi.org/10.3857/roj.2023.01067] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38946077]
52. Hickey, B.A.; Chalmers, T.; Newton, P.; Lin, C.T.; Sibbritt, D.; McLachlan, C.S.; Clifton-Bligh, R.; Morley, J.; Lal, S. Smart devices and wearable technologies to detect and monitor mental health conditions and stress: A systematic review. Sensors; 2021; 21, 3461. [DOI: https://dx.doi.org/10.3390/s21103461] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34065620]
53. Chalmers, T.; Hickey, B.A.; Newton, P.; Lin, C.T.; Sibbritt, D.; McLachlan, C.S.; Clifton-Bligh, R.; Morley, J.; Lal, S. Stress watch: The use of heart rate and heart rate variability to detect stress: A pilot study using smart watch wearables. Sensors; 2021; 22, 151. [DOI: https://dx.doi.org/10.3390/s22010151] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35009696]
54. Kang, M.; Chai, K. Wearable sensing systems for monitoring mental health. Sensors; 2022; 22, 994. [DOI: https://dx.doi.org/10.3390/s22030994]
55. Machado-Jaimes, L.G.; Bustamante-Bello, M.R.; Argüelles-Cruz, A.J.; Alfaro-Ponce, M. Development of an Intelligent System for the Monitoring and Diagnosis of the Well-Being. Sensors; 2022; 22, 9719. [DOI: https://dx.doi.org/10.3390/s22249719]
56. Gomes, N.; Pato, M.; Lourenco, A.R.; Datia, N. A survey on wearable sensors for mental health monitoring. Sensors; 2023; 23, 1330. [DOI: https://dx.doi.org/10.3390/s23031330]
57. Rohani, R. Meningkatkan Minat Belajar Siswa pada Pelajaran Matematika di Kelas XI Mia-3 SMA Negeri 2 Lubuk Pakam melalui Layanan Bimbingan Kelompok Tahun Pelajaran 2019/2020. J. Guru Kita PGSD; 2020; 4, [DOI: https://dx.doi.org/10.24114/jgk.v4i2.21928]
58. Gjoreski, M.; Luštrek, M.; Gams, M.; Gjoreski, H. Monitoring stress with a wrist device using context. J. Biomed. Inform.; 2017; 73, pp. 159-170. [DOI: https://dx.doi.org/10.1016/j.jbi.2017.08.006]
59. Ghosh, A.; Munoz-Munoz, N.; Lacoste, D. Minimum ignition energy of hydrogen-air and methane-air mixtures at temperatures as low as 200 K. Int. J. Hydrog. Energy; 2022; 47, pp. 30653-30659. [DOI: https://dx.doi.org/10.1016/j.ijhydene.2022.07.017]
60. General Data Protection Regulation (GDPR). 2024; Available online: https://eur-lex.europa.eu/eli/reg/2016/679/oj (accessed on 24 June 2025).
© 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Details
; Sideraki Angeliki
; Anagnostopoulos Christos-Nikolaos