The Role of Artificial Intelligence in Predicting and Managing Pediatric Fracture Overgrowth: A Comprehensive Review

Abstract

Pediatric fracture overgrowth is an unpredictable complication of long bone fractures in children, leading to excessive growth of the injured limb and resulting in limb length discrepancies (LLDs) and angular deformities that impact mobility and quality of life. Traditional methods struggle to predict at-risk children, hindering early detection and management. Artificial intelligence (AI), including machine learning and deep learning, offers advanced data analysis capabilities to enhance predictive accuracy and personalize treatment strategies. This comprehensive review explores the current understanding of pediatric fracture overgrowth, examines AI applications in medicine and orthopedics, evaluates potential AI applications specific to fracture overgrowth, and discusses ethical considerations and patient-centric approaches. We highlight how AI can improve diagnostic precision, facilitate early intervention, and optimize clinical outcomes. Though direct studies on AI in fracture overgrowth are limited, evidence from related areas underscores its potential. Embracing AI could revolutionize pediatric fracture management, leading to earlier detection, targeted interventions, and better outcomes for affected children.

1. Introduction

Fractures are a common occurrence in childhood due to high levels of physical activity and ongoing skeletal development. Children engage in activities such as climbing, running, and sports, increasing the risk of injuries including fractures. Epidemiological studies indicate that approximately 42% of boys and 27% of girls sustain at least one fracture before the age of 16 [1]. These injuries present unique challenges to healthcare providers due to the presence of growth plates (physes), which are areas of developing cartilage tissue near the ends of long bones. Growth plates are crucial for longitudinal bone growth but also introduce complexities in fracture healing and potential complications. One significant complication that can arise following long bone fractures in children is fracture overgrowth [2]. Unlike adults, where fractures typically heal to restore the bone’s original length and structure, children’s bones can exhibit accelerated growth during the healing process. This accelerated growth can lead to the injured limb becoming longer than its uninjured counterpart, resulting in limb length discrepancies (LLDs) and angular deformities. These discrepancies can have profound effects on a child’s gait, posture, and overall quality of life, potentially leading to functional impairments and psychological distress [3]. The unpredictable nature of fracture overgrowth adds complexity to pediatric fracture management. Traditional predictors, such as the child’s age, the type of fracture, and its location, provide only general guidance and often lack the precision needed for individualized risk assessment. This unpredictability makes early detection and effective management challenging, and in many cases, corrective surgeries such as epiphysiodesis or limb lengthening procedures become necessary. These interventions carry inherent risks and can significantly impact the child’s life, both physically and emotionally. In the face of these challenges, there is a growing interest in leveraging advanced technologies to improve outcomes in pediatric fracture care. Artificial intelligence (AI), encompassing machine learning (ML) and deep learning (DL), has emerged as a powerful tool in healthcare. AI systems can analyze vast amounts of complex data, identifying patterns and making predictions that may not be apparent through traditional analysis. In fields such as radiology, pathology, and adult orthopedics, AI has transformed predictive modeling and personalized treatment strategies. Supervised learning algorithms, including random forests, support vector machines, and neural networks, have been successfully applied to predict outcomes and complications in various medical conditions [4,5,6]. Given the potential of AI to revolutionize various aspects of medicine, its application in predicting and managing pediatric fracture overgrowth is a logical and promising next step. By integrating clinical data, imaging findings, genetic information, and treatment details, AI could provide more accurate risk assessments and facilitate early interventions tailored to individual patients. This could not only improve clinical outcomes but also reduce the burden on healthcare systems and families [7].

This comprehensive review aims to explore the intersection of AI and pediatric fracture overgrowth. We will delve into the current understanding of fracture overgrowth, examining its epidemiology, pathophysiology, and clinical implications. We will discuss the emergence of AI in medicine and orthopedics, highlighting key technologies and applications. Furthermore, we will examine existing evidence and potential applications of AI specific to fracture overgrowth and discuss ethical considerations, challenges, and future research directions. By synthesizing the existing literature, we hope to shed light on how AI can transform the management of pediatric fractures and stimulate further research and innovation in this critical area.

2. Understanding Pediatric Fracture Overgrowth

Fracture overgrowth in children is a phenomenon characterized by the injured bone exhibiting accelerated growth during the healing process, surpassing the length of the uninjured limb [2]. This overgrowth can lead to significant limb length discrepancies (LLDs) and angular deformities, which have lasting impacts on a child’s physical function and quality of life. The incidence of fracture overgrowth varies widely in the literature, reflecting the multifactorial nature of this condition. Reports suggest that overgrowth can occur in anywhere from 8% to 41% of pediatric femoral fractures [2,8]. Younger children are at higher risk, particularly those under the age of 10, due to their active growth plates and greater remaining growth potential [9,10]. The femur and tibia are the bones most commonly associated with overgrowth, possibly because of their significant contribution to overall limb length [11,12].

2.1. Mechanisms of Fracture Overgrowth

The mechanisms behind fracture overgrowth are not fully understood, but several theories have been proposed:

Hyperemia Theory: Following a fracture, increased blood flow (hyperemia) to the injury site facilitates healing. This increased vascularity may extend to the adjacent growth plate (physis), stimulating it to produce more cartilage cells that later ossify, thus accelerating longitudinal bone growth [13]. Hyperemia not only aids in fracture healing but may also inadvertently enhance growth plate activity.

Mechanical Environment: Immobilization of the fractured limb reduces mechanical stress on the growth plate, potentially altering the normal inhibitory signals that regulate growth plate activity and leading to an increase in growth rate [14,15]. Changes in muscle forces due to immobilization or altered weight-bearing can also impact growth plate stimulation.

Biological Factors: The fracture healing process involves a complex cascade of biological events, including the release of growth factors and cytokines. Molecules such as insulin-like growth factor-1 (IGF-1), bone morphogenetic proteins (BMPs), and vascular endothelial growth factor (VEGF) play crucial roles in bone repair and regeneration [13]. These factors may have systemic effects, enhancing growth plate activity beyond the immediate vicinity of the fracture.

Neurovascular Mechanisms: Injury to nerves and blood vessels during a fracture could disrupt normal growth regulation. Neurovascular mechanisms influence bone growth, and damage to these systems may lead to altered signaling at the growth plate, promoting overgrowth [16].

Genetic Factors: Emerging research suggests that genetic factors may also play a role. Variations in genes that regulate growth plate activity, bone healing, and response to injury could make some children more susceptible to overgrowth [17]. Understanding these genetic influences could lead to the identification of biomarkers for risk assessment [18].

2.2. Clinical Implications

The consequences of fracture overgrowth extend beyond physical discrepancies. Limb length discrepancies can lead to gait abnormalities, musculoskeletal strain, and functional limitations [19]. Children may develop abnormal gait patterns to compensate for the difference in limb length, such as limping or toe-walking. These compensatory mechanisms place additional stress on muscles, joints, and the spine, potentially leading to pain and discomfort. Activities requiring balance and coordination may be affected, impacting the child’s participation in sports and daily activities [20]. When overgrowth occurs asymmetrically across the growth plate, angular deformities like valgus (knock-knees) or varus (bow-legs) can develop. These deformities affect joint alignment and load distribution, increasing the risk of joint degeneration and osteoarthritis in the long term [3]. The psychological impact on children should not be underestimated. Visible limb discrepancies or deformities may lead to self-esteem issues, social withdrawal, and decreased participation in activities [21].

2.3. Challenges in Prediction and Management

Predicting which children will develop fracture overgrowth remains a significant challenge. Traditional predictors offer only general guidance and lack specificity. Early detection requires careful monitoring over time, but subtle discrepancies may not be clinically apparent until they become more significant [22]. Management options include observation, orthotic devices, and surgical interventions like epiphysiodesis (surgical arrest of the growth plate) or limb lengthening [23]. These interventions carry risks, including surgical complications and psychological effects from multiple procedures [24].

3. Artificial Intelligence in Medicine and Orthopedics

Artificial intelligence has marked a paradigm shift in healthcare, moving from a futuristic concept to a tangible reality reshaping clinical practice. AI refers to the ability of machines to mimic human cognitive functions such as learning and problem-solving. In medicine, AI’s capacity to process and analyze vast amounts of complex data far surpasses that of traditional statistical methods, allowing for the extraction of subtle patterns and insights that can enhance diagnostic accuracy, prognostic assessments, and therapeutic interventions [25].

3.1. Machine Learning and Deep Learning Techniques

At the heart of AI are technologies such as machine learning and deep learning, which have shown remarkable prowess in handling both structured and unstructured data [26]. Machine learning involves the development of algorithms that can learn from and make predictions based on data. It operates by identifying patterns within datasets and using these patterns to make decisions or predictions when presented with new data. This capability is particularly useful in processing structured data, such as numerical values from laboratory results, vital signs, and patient demographics [27].

Supervised Learning Algorithms: Algorithms like decision trees, random forests, support vector machines, and gradient boosting machines (e.g., XGBoost) have been widely used in healthcare for predictive modeling [4,5,6]. These algorithms require labeled data to learn the mapping between input features and output targets, making them suitable for classification and regression tasks.

Deep Learning: A subset of machine learning, deep learning utilizes artificial neural networks inspired by the human brain’s structure and function. These networks consist of multiple layers that can learn representations of data with multiple levels of abstraction. Convolutional neural networks (CNNs) are particularly effective in image recognition tasks and have been applied extensively in medical imaging [25]. Recurrent neural networks (RNNs) and long short-term memory networks (LSTMs) are used for sequential data, such as time-series analysis in patient monitoring [28].

Natural Language Processing (NLP): NLP enables computers to interpret, understand, and generate human language in a meaningful way. In the clinical context, NLP can be used to analyze unstructured text from electronic health records (EHRs), extracting valuable information from clinical notes, discharge summaries, and patient narratives [29]. This information can then be used to identify patient cohorts, predict outcomes, or support clinical decision-making.

3.2. AI in Orthopedic Practice

In orthopedics, AI applications are expanding rapidly:

Imaging Analysis: Deep learning algorithms can detect and classify fractures on radiographs with remarkable accuracy, sometimes surpassing that of human experts [30]. For instance, CNNs have been used to identify wrist fractures on radiographs, achieving performance comparable to experienced radiologists [31]. Shelmerdine et al. [32] conducted a systematic review highlighting AI’s role in radiological pediatric fracture assessment, emphasizing its potential to improve diagnostic accuracy and reduce interpretation times.

Automated Bone Age Assessment: Particularly relevant to pediatrics, AI-driven tools can automate bone age assessment using hand radiographs, providing rapid and consistent evaluations crucial for managing growth disorders [33].

Surgical Planning and Intraoperative Guidance: AI algorithms assist in selecting appropriate implants, determining optimal surgical approaches, and simulating surgical outcomes [34]. Intraoperatively, AI can enhance robotic-assisted surgeries by providing real-time feedback and adjustments.

Postoperative Care and Rehabilitation: Wearable devices and mobile applications equipped with AI monitor patients’ progress, adherence to rehabilitation protocols, and detect early signs of complications. By analyzing data from sensors tracking movement, strength, and range of motion, AI provides personalized feedback and adjusts rehabilitation programs [35].

The potential for AI to predict pediatric fracture overgrowth specifically is significant, though it remains a relatively unexplored area. Integrating diverse data sources—including detailed clinical information, imaging findings, genetic markers, and treatment modalities—could enable the development of comprehensive predictive models. Machine learning algorithms could analyze these complex data to identify patterns and risk factors associated with overgrowth, which may not be evident through traditional analysis.

4. Current Evidence and Applications

The application of artificial intelligence in orthopedics has been steadily growing, with research demonstrating its potential to enhance various aspects of clinical practice. While specific studies focusing on AI’s role in predicting pediatric fracture overgrowth are limited, the existing literature in related areas provides valuable insights into how AI can be leveraged to improve outcomes in pediatric orthopedics.

4.1. Predictive Modeling in Orthopedics

One of the key areas where AI has shown promise is in predictive modeling for orthopedic outcomes. Machine learning algorithms have been used to predict complications such as nonunion, infection, and the need for surgical intervention. In the context of fracture healing, the Machine Learning Consortium developed a machine learning model to predict nonunion in tibial shaft fractures by analyzing clinical data and surgical notes [36]. The model incorporated variables such as patient demographics, comorbidities, fracture characteristics, and intraoperative details. By processing both structured data and unstructured text through natural language processing techniques, the model achieved an accuracy of 85%. This study highlights the potential of AI to analyze complex and heterogeneous data sources to predict fracture outcomes. Similarly, in pediatric fracture management, Yao et al. applied machine learning algorithms to predict the necessity of surgical intervention in children with supracondylar humerus fractures [37]. By considering factors like patient age, mechanism of injury, physical examination findings, and initial radiographic measurements, the model provided clinicians with a decision support tool to aid in treatment planning. Such predictive models can help optimize resource utilization, reduce unnecessary interventions, and improve patient care. In other medical fields, AI has been successfully applied to predict outcomes in conditions like coronary artery disease and COVID-19 [38], demonstrating the transferability of methods and reinforcing the potential in orthopedics.

4.2. Deep Learning in Imaging Analysis

Deep learning has revolutionized imaging analysis in orthopedics, with algorithms capable of detecting and classifying fractures, assessing bone quality, and evaluating joint health. Lindsey et al. reported that a deep learning algorithm could detect and classify wrist fractures on radiographs with performance comparable to expert radiologists [39]. The algorithm utilized convolutional neural networks trained on a large dataset of annotated images, enabling it to identify subtle features indicative of fractures. The implementation of such tools in clinical practice can enhance diagnostic accuracy, reduce interpretation times, and support less experienced clinicians in making accurate diagnoses. In pediatric patients, the assessment of growth plates and bone development is critical. Lee et al. developed a deep learning system to automate the evaluation of bone age using hand radiographs [40]. The system demonstrated high accuracy and consistency compared to manual assessments by pediatric radiologists. Automated bone age assessment can streamline workflows, reduce variability, and provide objective measurements essential for managing growth disorders and planning treatments that may affect skeletal development. Furthermore, AI algorithms have been explored for analyzing growth plates to detect abnormalities that could contribute to complications like fracture overgrowth. By segmenting and quantifying features of the growth plate on imaging studies, AI can assist in identifying early signs of altered growth activity, potentially guiding interventions to prevent overgrowth.

4.3. Personalized Treatment Strategies

Artificial intelligence can enhance personalized medicine in orthopedics by providing risk stratification tools and optimizing treatment strategies based on individual patient characteristics. In the management of scoliosis, machine learning models have been developed to predict curve progression, aiding in decision-making regarding bracing or surgical intervention [41]. By analyzing factors such as age, sex, initial curve magnitude, and skeletal maturity, these models can identify patients at higher risk of progression, allowing for timely and appropriate interventions. Similarly, AI has been used to predict the risk of postoperative infections in orthopedic surgeries. By incorporating patient comorbidities, laboratory values, surgical factors, and intraoperative data, machine learning models can identify patients at increased risk, prompting the implementation of preventive measures [42]. In pediatric fracture care, predictive models can assist in determining the most appropriate treatment modality. For example, in forearm fractures, machine learning algorithms have been used to predict the likelihood of successful conservative management versus the need for surgical intervention [43]. By considering factors such as fracture displacement, angulation, and patient age, clinicians can make informed decisions that optimize outcomes and minimize unnecessary procedures.

4.4. Potential Applications in Fracture Overgrowth

Applying AI to predict fracture overgrowth involves the following (Table 1):

Predictive Modeling: Developing ML models incorporating fracture characteristics, patient-specific factors like age, sex, genetic markers, and hormonal influences to enhance predictive power [44].

Imaging Analysis: Utilizing DL algorithms to analyze imaging studies for early signs of altered growth plate activity, such as changes in thickness or signal intensity on MRI [45].

Monitoring Tools: Implementing AI-driven applications or wearable devices that monitor limb growth over time, detecting deviations from expected patterns [46].

Treatment Optimization: AI assists in selecting appropriate intervention strategies based on predicted risk, guiding clinicians in tailoring monitoring and interventions [36].

5. Ethical Considerations and Patient-Centric Focus

The integration of artificial intelligence into pediatric fracture management brings forth a range of ethical considerations that must be carefully addressed to ensure the responsible and equitable use of these technologies. These considerations encompass data privacy and security, informed consent, algorithmic bias, transparency, and the impact on the clinician–patient relationship.

5.1. Data Privacy and Security

In the era of big data and AI, the protection of patient information is paramount. Pediatric data are especially sensitive due to the vulnerability of the population and the potential long-term implications of data breaches [47]. Regulations such as the Health Insurance Portability and Accountability Act (HIPAA) in the United States and the General Data Protection Regulation (GDPR) in the European Union set stringent standards for data privacy. Collecting and utilizing pediatric health data for AI applications require robust measures to ensure data security [48]. This includes the encryption of data during transmission and storage, access controls to limit data handling to authorized personnel, and regular audits to detect and prevent unauthorized access. Data de-identification or anonymization is essential to protect patient identities, but care must be taken to prevent re-identification through data triangulation [49].

5.2. Informed Consent and Assent

Obtaining informed consent in pediatric populations involves both the legal guardians and, when appropriate, the child’s assent. The use of AI adds complexity to the consent process, as patients and guardians may have limited understanding of how AI works and its implications [50]. Clinicians and researchers must provide clear and accessible explanations of how patient data will be used in AI applications, including potential risks and benefits. This may involve educational materials or consultations with specialists in ethics or patient education. Transparency about data usage, storage duration, and the possibility of data sharing with third parties is crucial.

5.3. Algorithmic Bias and Equity

Algorithmic bias occurs when AI models produce systematically unfair outcomes due to biases in the training data or the algorithm itself. In healthcare, this can lead to disparities in diagnosis, treatment recommendations, or risk predictions among different demographic groups [51]. To prevent bias, AI models should be trained on diverse datasets that represent the population’s variability in terms of age, sex, race, ethnicity, socioeconomic status, and health conditions. Regular evaluation and validation of models across different subgroups can help identify and correct biases. Involving stakeholders from diverse backgrounds in the development and testing of AI applications can also enhance fairness. Huang et al. [52] emphasized transparency through methods like SHAP values to interpret model predictions and identify feature importance.

5.4. Transparency and Explainability

For AI tools to be trusted and adopted in clinical practice, they must be transparent and explainable. Clinicians need to understand how AI models arrive at their predictions or recommendations to assess their validity and integrate them into clinical decision-making [53]. Explainable AI (XAI) aims to make AI models’ inner workings understandable to humans. Techniques such as model interpretability, the visualization of decision pathways, and the use of simpler algorithms where appropriate can enhance transparency [54,55]. Providing clinicians with explanations of the key factors influencing AI outputs allows them to evaluate the relevance and reliability of the information.

5.5. Impact on the Clinician–Patient Relationship

The introduction of AI into clinical care should enhance, not hinder, the clinician–patient relationship. There is concern that reliance on AI could depersonalize care or reduce the clinician’s engagement with the patient. To mitigate this risk, AI should be viewed as a tool that supports clinicians rather than replaces them. Emphasizing the clinician’s role in interpreting AI outputs, contextualizing them within the patient’s broader clinical picture, and communicating findings to patients and families are essential. Clinicians should maintain their responsibility for final decisions and ensure that AI complements their expertise [56].

5.6. Legal and Regulatory Considerations

The deployment of AI in healthcare is subject to regulatory oversight to ensure safety and efficacy. Regulatory bodies like the U.S. Food and Drug Administration (FDA) have begun developing frameworks for the evaluation of AI-based medical devices and software [57]. Compliance with regulations requires thorough validation of AI models, documentation of development processes, and post-market surveillance to monitor performance and detect adverse events. Engaging with regulators early in the development process can facilitate compliance and expedite the approval process.

5.7. Ethical Use of AI in Pediatrics

Ethical frameworks specific to pediatric AI applications are needed to guide practitioners and developers. These frameworks should address issues such as the following:

Beneficence and Non-Maleficence: Ensuring that AI applications provide a net benefit to patients and do not cause harm, by including rigorous testing and validation before clinical use;

Autonomy: Respecting the rights of patients and families to make informed decisions about their care, including the use of AI tools;

Justice: Promoting equitable access to AI technologies and preventing disparities in care;

Accountability: Establishing clear lines of responsibility for AI decisions and ensuring mechanisms are in place for addressing errors or adverse outcomes.

5.8. Stakeholder Engagement

Involving patients, families, clinicians, ethicists, and other stakeholders in the development and implementation of AI applications can enhance their alignment with patient needs and values. Patient advocacy groups can provide insights into patient priorities, concerns, and preferences. Public education about AI in healthcare is also important. Providing accessible information about the benefits and risks of AI can build trust and facilitate acceptance. Open communication about how AI is used in patient care, its success, and its limitations can foster transparency.

6. Challenges and Limitations

Despite the promising potential of artificial intelligence in predicting and managing pediatric fracture overgrowth, several challenges and limitations must be addressed to fully realize its benefits. These challenges span technical, clinical, ethical, and practical domains.

6.1. Data Availability and Quality

One of the primary challenges in developing AI models for pediatric fracture overgrowth is the availability of high-quality, comprehensive datasets. Pediatric orthopedic cases are less common than adult cases, resulting in smaller datasets. Additionally, ethical considerations and regulations around pediatric data limit the sharing and use of patient information. Data heterogeneity further complicates model development. Variations in imaging protocols, equipment, and settings across different institutions lead to inconsistencies in image quality and characteristics. Differences in clinical documentation, terminology, and treatment approaches introduce additional variability [58]. To overcome these challenges, multicenter collaborations are essential. By pooling data from multiple institutions, researchers can increase dataset size and diversity, enhancing model robustness and generalizability. Standardization efforts, such as developing uniform imaging protocols and data collection practices, can improve data consistency. Data augmentation techniques, including the use of synthetic data and image augmentation, can expand datasets [59]. However, care must be taken to ensure that augmented data accurately represent real-world variations and do not introduce biases.

6.2. Model Generalizability and Bias

AI models trained on specific datasets may not perform well when applied to new populations or settings. This lack of generalizability can result from overfitting to the training data or from biases present in the data. Algorithmic bias is a significant concern, particularly in pediatric populations where variability in growth and development is high [60,61]. If models are trained predominantly on data from certain demographic groups, they may not perform adequately for others, leading to disparities in care. Mitigation strategies include the following:

Diverse Training Data: Ensuring that datasets represent a wide range of patient demographics, clinical presentations, and imaging characteristics;

Cross-Validation: Using techniques like k-fold cross-validation to assess model performance on different subsets of data;

External Validation: Testing models on independent datasets from other institutions or populations to evaluate generalizability;

Bias Detection and Correction: Implementing methods to identify biases in model outputs and adjusting algorithms accordingly.

6.3. Integration into Clinical Practice

Implementing AI tools in clinical practice presents both technical and cultural challenges. Clinicians may be skeptical of AI technologies, particularly if they perceive them as black boxes with opaque decision-making processes [62]. Concerns about job displacement or loss of professional autonomy may also hinder acceptance. From a technical standpoint, integrating AI tools with existing electronic health record (EHR) systems and workflows requires significant effort [63]. Compatibility issues, data interoperability, and user interface design are critical factors that influence usability.

To facilitate adoption, the following should be considered:

Education and Training: Providing clinicians with education about AI technologies, their capabilities, their limitations, and how they can enhance patient care;

User-Centered Design: Involving clinicians in the design and development of AI tools to ensure they meet clinical needs and integrate smoothly into workflows;

Transparency and Explainability: Ensuring that AI models are transparent and that clinicians can understand and interpret their outputs;

Demonstrating Value: Conducting studies that demonstrate the efficacy and cost-effectiveness of AI applications in improving patient outcomes.

6.4. Regulatory and Legal Challenges

Navigating the regulatory landscape for AI in healthcare is complex. AI applications may be considered medical devices or software as a medical device (SaMD), subjecting them to regulatory oversight. Regulations vary by country and may not be fully established for AI technologies. Compliance requires understanding regulatory requirements, conducting thorough validation and documentation, and engaging with regulatory agencies. Legal considerations also include liability issues [64]. Determining responsibility in cases where AI contributes to an adverse event is a challenge. Clear guidelines and legal frameworks are needed to address liability and ensure accountability.

6.5. Ethical and Social Implications

The use of AI in pediatrics raises ethical questions related to autonomy, consent, and the potential impact on the clinician–patient relationship. There is a risk that over-reliance on AI could depersonalize care or undermine the clinician’s role. Social implications include the potential for AI to exacerbate disparities if access to technology is unequal [65]. Ensuring equitable access to AI tools, particularly in underserved or resource-limited settings, is essential. Public perception of AI in healthcare may be influenced by misinformation or unrealistic expectations. Engaging with the public to provide accurate information and address concerns is important for building trust.

7. Future Research Directions and Implementation Strategies

To fully harness the potential of artificial intelligence in predicting and managing pediatric fracture overgrowth, concerted efforts in research, development, and implementation are required. Future directions involve advancing predictive models, exploring novel AI techniques, fostering interdisciplinary collaboration, and addressing ethical and legal considerations. Advancing predictive models specifically tailored to fracture overgrowth is a critical step forward. This endeavor requires comprehensive, high-quality datasets that encompass a wide range of variables influencing overgrowth risk. Prospective studies should be initiated to collect detailed clinical data, including patient demographics, medical history, fracture characteristics, treatment methods, and outcomes [66]. Imaging data should be standardized, capturing high-resolution images of fractures and growth plates using consistent protocols. Incorporating genetic and molecular data, such as genomic markers associated with growth plate activity, could provide deeper insights into individual susceptibility. By integrating these diverse data sources, machine learning algorithms can be trained to identify complex patterns and interactions that contribute to overgrowth. Advanced modeling techniques, such as ensemble methods that combine multiple algorithms, can enhance predictive accuracy. Temporal modeling using recurrent neural networks (RNNs) can capture the dynamic nature of growth and development, allowing predictions to adjust over time as new data become available [67]. Exploring novel AI techniques offers new avenues for research. Federated learning, for example, enables the development of AI models across multiple institutions without the need to share raw patient data. This approach enhances data diversity and model robustness while maintaining patient privacy. Explainable AI (XAI) techniques are essential for building trust and facilitating clinical adoption [55]. By developing models that provide clear explanations of their decision-making processes, clinicians can better understand and interpret AI outputs, integrating them effectively into patient care. Interdisciplinary collaboration is essential to address the multifaceted challenges of applying AI to pediatric fracture overgrowth. Data scientists and AI researchers bring technical expertise in algorithm development and data analysis. Orthopedic surgeons and pediatricians provide clinical insights and ensure that models address relevant clinical questions. Ethicists play a crucial role in guiding the responsible use of AI, addressing concerns related to privacy, consent, and equity. Engaging patients and families in the research process ensures that patient perspectives and needs are considered, enhancing the relevance and acceptability of AI applications [68]. Implementing pilot programs in select clinical settings allows for the assessment of feasibility, usability, and impact on patient care. These programs provide opportunities to refine algorithms based on real-world feedback and identify practical challenges in integration. Establishing feedback mechanisms is essential. Clinicians using AI tools should have channels to report issues, suggest improvements, and share experiences. Continuous monitoring of AI performance in the clinical environment ensures that models remain accurate and effective over time. Developing ethical guidelines specific to the use of AI in pediatric care is imperative. These guidelines should address issues such as informed consent, data privacy, algorithmic transparency, and the prevention of bias [68]. Collaboration with regulatory bodies, such as the FDA, can help ensure that AI applications comply with legal requirements and maintain high standards of safety and efficacy. Engaging with patient advocacy groups and the broader public is important for building trust and acceptance. Public education initiatives can inform patients and families about the benefits and risks of AI, addressing misconceptions and fostering informed decision-making [69]. Ensuring equitable access to AI technologies is a key consideration. Efforts should be made to provide resources and support to underserved communities, including rural areas and low-income populations. This may involve developing cost-effective solutions, leveraging telemedicine, and providing training and infrastructure support. Research should also focus on evaluating the impact of AI applications on health disparities. By monitoring outcomes across different demographic groups, interventions can be adjusted to promote equity and prevent the exacerbation of existing disparities [70]. The field of AI is rapidly evolving, and continuous evaluation is necessary to keep pace with new developments. Regular assessments of AI models’ performance, including their accuracy, reliability, and impact on patient outcomes, are essential. The adaptation of models to incorporate new data, address emerging challenges, and integrate novel technologies ensures that AI applications remain relevant and effective [71]. Engaging in ongoing research and staying informed about advances in the field enable clinicians and researchers to leverage AI’s full potential in improving pediatric fracture care.

8. Conclusions

Pediatric fracture overgrowth remains a challenging complication due to its unpredictability and significant impact on a child’s life. Traditional methods are limited, often leading to delayed interventions. Artificial intelligence offers a promising avenue for enhancing predictive capabilities and personalize treatment. While direct applications of AI in fracture overgrowth are developing, success in related areas highlights its potential. AI technologies can process complex data to identify patterns not apparent through traditional analysis. Overcoming challenges related to data availability, bias, ethical considerations, and clinical integration is essential. Embracing AI represents a significant step toward precision medicine in pediatrics. Collaborative efforts, including multicenter studies and international collaborations, are needed to validate predictive models and facilitate integration into practice. AI holds the promise of revolutionizing care and improving outcomes for children affected by fracture overgrowth.

Author Contributions

Conceptualization, D.A.M. and B.Z.; methodology, M.N.; formal analysis, M.N. and I.S.; investigation, D.A.M.; writing original draft preparation, G.G. and D.A.M.; writing review and editing, G.G.; supervision, D.L. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

No new data were created or analyzed in this study.

Conflicts of Interest

The authors declare no conflicts of interest.

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.

Table

Table 1

Comparison of traditional and AI-driven approaches to pediatric fracture overgrowth management.

	Traditional Approach	AI-Driven Approach
Predictive Accuracy	Low; relies on clinical judgment	High; utilizes data-driven models
Early Detection	Challenging; often reactive	Facilitated; proactive identification
Personalized Treatment	Limited; generalized protocols	Enhanced; tailored interventions
Resource Utilization	Variable; may lead to over-/under-treatment	Optimized; efficient allocation
Clinician Workload	High; manual assessments	Reduced; automated analysis
Patient Outcomes	Variable; risk of complications	Improved; potential reduction in LLDs/deformities

References

1. Hedström, E.M.; Svensson, O.; Bergström, U.; Michno, P. Epidemiology of fractures in children and adolescents. Acta Orthop.; 2010; 81, pp. 148-153. [DOI: https://dx.doi.org/10.3109/17453671003628780] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20175744]

2. Shapiro, F. Fractures of the Femoral Shaft in Children: The Overgrowth Phenomenon. Acta Orthop. Scand.; 1981; 52, pp. 649-655. [DOI: https://dx.doi.org/10.3109/17453678108992162]

3. Kim, T.G.; Park, M.S.; Lee, S.H.; Choi, K.J.; Im, B.-E.; Kim, D.Y.; Sung, K.H. Leg-length discrepancy and associated risk factors after paediatric femur shaft fracture: A multicentre study. J. Child. Orthop.; 2021; 15, pp. 215-222. [DOI: https://dx.doi.org/10.1302/1863-2548.15.200252] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34211597]

4. Breiman, L. Random forests. Mach. Learn.; 2001; 45, pp. 5-32. [DOI: https://dx.doi.org/10.1023/A:1010933404324]

5. Chen, T.; Guestrin, C. XGBoost: A Scalable Tree Boosting System. Proceedings of the KDD ’16: 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining; San Francisco, CA, USA, 13–17 August 2016; Association for Computing Machinery: New York, NY, USA, 2016; pp. 785-794. [DOI: https://dx.doi.org/10.1145/2939672.2939785]

6. Cortes, C.; Vapnik, V. Support-vector networks. Mach. Learn.; 1995; 20, pp. 273-297. [DOI: https://dx.doi.org/10.1007/BF00994018]

7. Topol, E.J. High-performance medicine: The convergence of human and artificial intelligence. Nat. Med.; 2019; 25, pp. 44-56. [DOI: https://dx.doi.org/10.1038/s41591-018-0300-7]

8. Edvardsen, P.; Syversen, S. Overgrowth of the femur after fracture of the shaft in childhood. J. Bone Jt. Surg. Br. Vol.; 1976; 58-B, pp. 339-342. [DOI: https://dx.doi.org/10.1302/0301-620X.58B3.956252] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/956252]

9. Wenger, D.R. Practice in Paediatric Orthopaedics, Third Edition. J. Pediatr. Orthop.; 2017; 37, e302. [DOI: https://dx.doi.org/10.1097/BPO.0000000000000885]

10. Kuo, F.-C.; Kuo, S.-J.; Ko, J.-Y. Overgrowth of the femoral neck after hip fractures in children. J. Orthop. Surg. Res.; 2016; 11, 50. [DOI: https://dx.doi.org/10.1186/s13018-016-0387-9]

11. Buckley, S.L. Current trends in the treatment of femoral shaft fractures in children and adolescents. Clin. Orthop. Relat. Res.; 1997; 338, pp. 60-73. [DOI: https://dx.doi.org/10.1097/00003086-199705000-00009] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/9170363]

12. Itonaga, I.; Iwasaki, T.; Kaku, N.; Tsumura, H. Bone growth after fracture of the femoral shaft in children 2023. Res. Sq.; 2023. [DOI: https://dx.doi.org/10.21203/rs.3.rs-3007342/v1]

13. May, C.; Spencer, S. The Pediatric Femur: Development, Growth, and Surgical Anatomy. Pediatric Femur Fractures; Springer: Boston, MA, USA, 2016; pp. 1-25. [DOI: https://dx.doi.org/10.1007/978-1-4899-7986-5_1]

14. Trueta, J. Blood supply and the rate of healing of tibial fractures. Clin. Orthop. Relat. Res.; 1974; 105, pp. 11-26. [DOI: https://dx.doi.org/10.1097/00003086-197411000-00003]

15. Etchebehere, E.C.; Caron, M.; Pereira, J.A.; Lima, M.C.; Santos, A.O.; Ramos, C.D.; Barros, F.B.; Sanches, A.; Santos-Jesus, R.; Belangero, W. et al. Activation of the growth plates on three-phase bone scintigraphy: The explanation for the overgrowth of fractured femurs. Eur. J. Nucl. Med.; 2001; 28, pp. 72-80. [DOI: https://dx.doi.org/10.1007/s002590000395] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/11202455]

16. Naik, P. Remodelling in Children’s Fractures and Limits of Acceptability. Indian J. Orthop.; 2021; 55, pp. 549-559. [DOI: https://dx.doi.org/10.1007/s43465-020-00320-2]

17. Baron, J.; O Klein, K.; Colli, M.J.; A Yanovski, J.; A Novosad, J.; Bacher, J.D.; Cutler, G.B. Catch-up growth after glucocorticoid excess: A mechanism intrinsic to the growth plate. Endocrinology; 1994; 135, pp. 1367-1371. [DOI: https://dx.doi.org/10.1210/endo.135.4.7925098]

18. Liu, C.; Liu, Y.; Zhang, W.; Liu, X. Screening for potential genes associated with bone overgrowth after mid-shaft femur fracture in a rat model. J. Orthop. Surg. Res.; 2017; 12, 8. [DOI: https://dx.doi.org/10.1186/s13018-017-0510-6]

19. Sanders, J.O.; Browne, R.H.; Mooney, J.F.; Raney, E.M.; Horn, B.D.; Anderson, D.J.; Hennrikus, W.L.; Robertson, W.W. Treatment of femoral fractures in children by pediatric orthopedists: Results of a 1998 survey. J. Pediatr. Orthop.; 2001; 21, pp. 436-441. [DOI: https://dx.doi.org/10.1097/01241398-200107000-00004]

20. Aiona, M.; Do, K.P.B.; Emara, K.; Dorociak, R.B.; Pierce, R.B. Gait Patterns in Children With Limb Length Discrepancy. J. Pediatr. Orthop.; 2015; 35, pp. 280-284. [DOI: https://dx.doi.org/10.1097/BPO.0000000000000262]

21. Toovey, R.; Coulston, F.; Shuttleworth, H.; Jelleff, M.; Villalon, J.; Anderson, S.; Shields, N.; Finnigan, M.; Lalor, S.; McGinley, J. Participation experiences of young people with limb difference in sports and recreation in Australia: A qualitative study. Disabil. Rehabil.; 2024; pp. 1-12. [DOI: https://dx.doi.org/10.1080/09638288.2024.2413180]

22. Burghardt, R.D.; von der Heydt, K.; Heckel, N.; Herzenberg, J.E.; Burgkart, R. Complications and Errors in Guided Growth Surgery. J. Limb Lengthening Reconstr.; 2023; 9, pp. 70-81. [DOI: https://dx.doi.org/10.4103/jllr.jllr_13_22]

23. Wright, J.G.; EL Wang, E.; Owen, J.L.; Stephens, D.; Graham, H.K.; Hanlon, M.; Nattrass, G.R.; Reynolds, R.A.; Coyte, P. Treatments for paediatric femoral fractures: A randomised trial. Lancet; 2005; 365, pp. 1153-1158. [DOI: https://dx.doi.org/10.1016/S0140-6736(05)71878-X] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15794970]

24. Bafor, A.; Iobst, C.A. What’s New in Limb Lengthening and Deformity Correction. J. Bone Jt. Surg.; 2023; 105, pp. 1221-1226. [DOI: https://dx.doi.org/10.2106/JBJS.23.00310] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37352371]

25. LeCun, Y.; Bengio, Y.; Hinton, G. Deep learning. Nature; 2015; 521, pp. 436-444. [DOI: https://dx.doi.org/10.1038/nature14539]

26. Esteva, A.; Kuprel, B.; Novoa, R.A.; Ko, J.; Swetter, S.M.; Blau, H.M.; Thrun, S. Dermatologist-level classification of skin cancer with deep neural networks. Nature; 2017; 542, pp. 115-118. [DOI: https://dx.doi.org/10.1038/nature21056] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28117445]

27. Deo, R.C. Machine Learning in Medicine. Circulation; 2015; 132, pp. 1920-1930. [DOI: https://dx.doi.org/10.1161/CIRCULATIONAHA.115.001593]

28. Men, L.; Ilk, N.; Tang, X.; Liu, Y. Multi-disease prediction using LSTM recurrent neural networks. Expert Syst. Appl.; 2021; 177, 114905. [DOI: https://dx.doi.org/10.1016/j.eswa.2021.114905]

29. Hossain, E.; Rana, R.; Higgins, N.; Soar, J.; Barua, P.D.; Pisani, A.R.; Turner, K. Natural Language Processing in Electronic Health Records in relation to healthcare decision-making: A systematic review. Comput. Biol. Med.; 2023; 155, 106649. [DOI: https://dx.doi.org/10.1016/j.compbiomed.2023.106649] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36805219]

30. Olczak, J.; Fahlberg, N.; Maki, A.; Razavian, A.S.; Jilert, A.; Stark, A.; Sköldenberg, O.; Gordon, M. Artificial intelligence for analyzing orthopedic trauma radiographs. Acta Orthop.; 2017; 88, pp. 581-586. [DOI: https://dx.doi.org/10.1080/17453674.2017.1344459]

31. Kim, D.; MacKinnon, T. Artificial intelligence in fracture detection: Transfer learning from deep convolutional neural networks. Clin. Radiol.; 2018; 73, pp. 439-445. [DOI: https://dx.doi.org/10.1016/j.crad.2017.11.015]

32. Shelmerdine, S.C.; White, R.D.; Liu, H.; Arthurs, O.J.; Sebire, N.J. Artificial intelligence for radiological paediatric fracture assessment: A systematic review. Insights Into Imaging; 2022; 13, 94. [DOI: https://dx.doi.org/10.1186/s13244-022-01234-3]

33. Larson, D.B.; Chen, M.C.; Lungren, M.P.; Halabi, S.S.; Stence, N.V.; Langlotz, C.P. Performance of a Deep-Learning Neural Network Model in Assessing Skeletal Maturity on Pediatric Hand Radiographs. Radiology; 2018; 287, pp. 313-322. [DOI: https://dx.doi.org/10.1148/radiol.2017170236] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29095675]

34. Wong, K.C.; Kumta, S.M. Computer-assisted tumor surgery in malignant bone tumors. Clin. Orthop. Relat. Res.; 2013; 471, pp. 750-761. [DOI: https://dx.doi.org/10.1007/s11999-012-2557-3] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22948530]

35. Dunn, J.; Runge, R.; Snyder, M. Wearables and the Medical Revolution. Pers. Med.; 2018; 15, pp. 429-448. [DOI: https://dx.doi.org/10.2217/pme-2018-0044] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30259801]

36. Bhandari, M.; Bulstra, A.E.J.; Bzovsky, S.; Doornberg, J.N.; Goslings, J.C.; Hendrickx, L.A.M.; Jaarsma, R.L.; Jeray, K.J.; Kerkhoffs, G.M.M.J.; Petrisor, B. et al. A Machine Learning Algorithm to Identify Patients with Tibial Shaft Fractures at Risk for Infection After Operative Treatment. J. Bone Jt. Surg.; 2021; 103, pp. 532-540. [DOI: https://dx.doi.org/10.2106/jbjs.20.00903]

37. Yao, W.M.; Wang, Y.; Zhao, X.M.; He, M.M.; Wang, Q.M.; Liu, H.M.; Zhao, J.M. Automatic diagnosis of pediatric supracondylar humerus fractures using radiomics-based machine learning. Medicine; 2024; 103, e38503. [DOI: https://dx.doi.org/10.1097/MD.0000000000038503] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38847664]

38. Payer, C.; Štern, D.; Bischof, H.; Urschler, M. Multi-label Whole Heart Segmentation Using CNNs and Anatomical Label Configurations. Statistical Atlases and Computational Models of the Heart; Springer: Cham, Switzerland, 2018; pp. 190-198. [DOI: https://dx.doi.org/10.1007/978-3-319-75541-0_20]

39. Lindsey, R.; Daluiski, A.; Chopra, S.; Lachapelle, A.; Mozer, M.; Sicular, S.; Hanel, D.; Gardner, M.; Gupta, A.; Hotchkiss, R. et al. Deep neural network improves fracture detection by clinicians. Proc. Natl. Acad. Sci. USA; 2018; 115, pp. 11591-11596. [DOI: https://dx.doi.org/10.1073/pnas.1806905115]

40. Lee, H.; Tajmir, S.; Lee, J.; Zissen, M.; Yeshiwas, B.A.; Alkasab, T.K.; Choy, G.; Do, S. Fully Automated Deep Learning System for Bone Age Assessment. J. Digit. Imaging; 2017; 30, pp. 427-441. [DOI: https://dx.doi.org/10.1007/s10278-017-9955-8]

41. Wang, M.X.; Kim, J.K.; Choi, J.-W.; Park, D.; Chang, M.C. Deep learning algorithm for automatically measuring Cobb angle in patients with idiopathic scoliosis. Eur. Spine J.; 2024; 33, pp. 4155-4163. [DOI: https://dx.doi.org/10.1007/s00586-023-08024-5] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38367024]

42. Karhade, A.V.; Thio, Q.C.B.S.; Ogink, P.T.; A Shah, A.; Bono, C.M.; Oh, K.S.; Saylor, P.J.; Schoenfeld, A.J.; Shin, J.H.; Harris, M.B. et al. Development of Machine Learning Algorithms for Prediction of 30-Day Mortality After Surgery for Spinal Metastasis. Neurosurgery; 2019; 85, pp. E83-E91. [DOI: https://dx.doi.org/10.1093/neuros/nyy469]

43. Mahan, S.T.; Katz, J.N.; Kim, Y.-J. To Screen or not to screen? A decision analysis of the utility of screening for developmental dysplasia of the hip. J. Bone Jt. Surg.; 2009; 91, pp. 1705-1719. [DOI: https://dx.doi.org/10.2106/JBJS.H.00122] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19571094]

44. Stahlschmidt, S.R.; Ulfenborg, B.; Synnergren, J. Multimodal deep learning for biomedical data fusion: A review. Briefings Bioinform.; 2022; 23, bbab569. [DOI: https://dx.doi.org/10.1093/bib/bbab569]

45. Malamateniou, C.; Knapp, K.; Pergola, M.; Woznitza, N.; Hardy, M. Artificial intelligence in radiography: Where are we now and what does the future hold?. Radiography; 2021; 27, (Suppl. 1), pp. S58-S62. [DOI: https://dx.doi.org/10.1016/j.radi.2021.07.015] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34380589]

46. Navaneethan, S.D.; Jolly, S.E.; Sharp, J.; Jain, A.; Schold, J.D.; Schreiber, M.J., Jr.; Nally, J.V., Jr. Electronic health records: A new tool to combat chronic kidney disease?. Clin. Nephrol.; 2013; 79, pp. 175-183. [DOI: https://dx.doi.org/10.5414/CN107757]

47. United Nations Publications. The Convention on the Rights of the Child; United Nations Publications: New York, NY, USA, 2019; [DOI: https://dx.doi.org/10.18356/f65bafb2-en]

48. Dove, E.S. The EU General Data Protection Regulation: Implications for International Scientific Research in the Digital Era. J. Law Med. Ethics; 2018; 46, pp. 1013-1030. [DOI: https://dx.doi.org/10.1177/1073110518822003]

49. Dwork, C.; Roth, A. The Algorithmic Foundations of Differential Privacy. Found. Trends Theor. Comput. Sci.; 2013; 9, pp. 211-407. [DOI: https://dx.doi.org/10.1561/0400000042]

50. Sheller, M.J.; Edwards, B.; Reina, G.A.; Martin, J.; Pati, S.; Kotrotsou, A.; Milchenko, M.; Xu, W.; Marcus, D.; Colen, R.R. et al. Federated learning in medicine: Facilitating multi-institutional collaborations without sharing patient data. Sci. Rep.; 2020; 10, 12598. [DOI: https://dx.doi.org/10.1038/s41598-020-69250-1] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32724046]

51. Li, T.; Sahu, A.K.; Talwalkar, A.; Smith, V. Federated Learning: Challenges, Methods, and Future Directions. IEEE Signal Process. Mag.; 2020; 37, pp. 50-60. [DOI: https://dx.doi.org/10.1109/MSP.2020.2975749]

52. Huang, A.A.; Huang, S.Y. Increasing transparency in machine learning through bootstrap simulation and shapely additive explanations. PLoS ONE; 2023; 18, e0281922. [DOI: https://dx.doi.org/10.1371/journal.pone.0281922] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36821544]

53. Obermeyer, Z.; Powers, B.; Vogeli, C.; Mullainathan, S. Dissecting racial bias in an algorithm used to manage the health of populations. Science; 2019; 366, pp. 447-453. [DOI: https://dx.doi.org/10.1126/science.aax2342]

54. Acharya, V.; Padhan, P.; Bahinipati, J.; Mishra, S.; Aggarwal, K.; Jhajharia, S.; Parida, P.; Sahu, D.; Pradhan, T. Artificial intelligence in medical education. J. Integr. Med. Res.; 2023; 1, 87. [DOI: https://dx.doi.org/10.4103/jimr.jimr_17_23]

55. Tjoa, E.; Guan, C. A Survey on Explainable Artificial Intelligence (XAI): Toward Medical XAI. IEEE Trans. Neural Networks Learn. Syst.; 2021; 32, pp. 4793-4813. [DOI: https://dx.doi.org/10.1109/TNNLS.2020.3027314] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33079674]

56. Shortliffe, E.H.; Sepúlveda, M.J. Clinical Decision Support in the Era of Artificial Intelligence. JAMA; 2018; 320, pp. 2199-2200. [DOI: https://dx.doi.org/10.1001/jama.2018.17163]

57. Amann, J.; Blasimme, A.; Vayena, E.; Frey, D.; Madai, V.I. Explainability for artificial intelligence in healthcare: A multidisciplinary perspective. BMC Med. Inform. Decis. Mak.; 2020; 20, 310. [DOI: https://dx.doi.org/10.1186/s12911-020-01332-6] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33256715]

58. Johnson, K.W.; Soto, J.T.; Glicksberg, B.S.; Shameer, K.; Miotto, R.; Ali, M.; Ashley, E.; Dudley, J.T. Artificial Intelligence in Cardiology. J. Am. Coll. Cardiol.; 2018; 71, pp. 2668-2679. [DOI: https://dx.doi.org/10.1016/j.jacc.2018.03.521]

59. Handelman, G.S.; Kok, H.K.; Chandra, R.V.; Razavi, A.H.; Lee, M.J.; Asadi, H. eDoctor: Machine learning and the future of medicine. J. Intern. Med.; 2018; 284, pp. 603-619. [DOI: https://dx.doi.org/10.1111/joim.12822]

60. Mehrabi, N.; Morstatter, F.; Saxena, N.; Lerman, K.; Galstyan, A. A Survey on Bias and Fairness in Machine Learning. ACM Comput. Surv.; 2022; 54, pp. 1-35. [DOI: https://dx.doi.org/10.1145/3457607]

61. Zhao, J.; Wang, T.; Yatskar, M.; Ordonez, V.; Chang, K.-W. Men Also Like Shopping: Reducing Gender Bias Amplification using Corpus-level Constraints. Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing; Copenhagen, Denmark, 9–11 September 2017; Association for Computational Linguistics: Stroudsburg, PA, USA, 2017; pp. 2979-2989. [DOI: https://dx.doi.org/10.18653/v1/D17-1323]

62. Sutton, R.T.; Pincock, D.; Baumgart, D.C.; Sadowski, D.C.; Fedorak, R.N.; Kroeker, K.I. An overview of clinical decision support systems: Benefits, risks, and strategies for success. npj Digit. Med.; 2020; 3, 17. [DOI: https://dx.doi.org/10.1038/s41746-020-0221-y]

63. Buch, V.H.; Ahmed, I.; Maruthappu, M. Artificial intelligence in medicine: Current trends and future possibilities. Br. J. Gen. Pract.; 2018; 68, pp. 143-144. [DOI: https://dx.doi.org/10.3399/bjgp18X695213]

64. Fraser, A.G.; Biasin, E.; Bijnens, B.; Bruining, N.; Caiani, E.G.; Cobbaert, K.; Davies, R.H.; Gilbert, S.H.; Hovestadt, L.; Kamenjasevic, E. et al. Artificial intelligence in medical device software and high-risk medical devices—A review of definitions, expert recommendations and regulatory initiatives. Expert Rev. Med. Devices; 2023; 20, pp. 467-491. [DOI: https://dx.doi.org/10.1080/17434440.2023.2184685] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37157833]

65. Bekbolatova, M.; Mayer, J.; Ong, C.W.; Toma, M. Transformative Potential of AI in Healthcare: Definitions, Applications, and Navigating the Ethical Landscape and Public Perspectives. Healthcare; 2024; 12, 125. [DOI: https://dx.doi.org/10.3390/healthcare12020125]

66. Chokesuwattanaskul, R.; Petchlorlian, A.; Lertsanguansinchai, P.; Suttirut, P.; Prasitlumkum, N.; Srimahachota, S.; Buddhari, W. Machine Learning-Based Predictive Model of Aortic Valve Replacement Modality Selection in Severe Aortic Stenosis Patients. Med. Sci.; 2023; 12, 3. [DOI: https://dx.doi.org/10.3390/medsci12010003] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38249079]

67. Choi, E.; Schuetz, A.; Stewart, W.F.; Sun, J. Using recurrent neural network models for early detection of heart failure onset. J. Am. Med. Inform. Assoc.; 2017; 24, pp. 361-370. [DOI: https://dx.doi.org/10.1093/jamia/ocw112] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27521897]

68. Rigby, M.J. Ethical Dimensions of Using Artificial Intelligence in Health Care. AMA J. Ethics; 2019; 21, pp. E121-E124. [DOI: https://dx.doi.org/10.1001/amajethics.2019.121]

69. Palaniappan, K.; Lin, E.Y.T.; Vogel, S. Global Regulatory Frameworks for the Use of Artificial Intelligence (AI) in the Healthcare Services Sector. Healthcare; 2024; 12, 562. [DOI: https://dx.doi.org/10.3390/healthcare12050562]

70. Fiske, A.; Henningsen, P.; Buyx, A. Your Robot Therapist Will See You Now: Ethical Implications of Embodied Artificial Intelligence in Psychiatry, Psychology, and Psychotherapy. J. Med. Internet Res.; 2019; 21, e13216. [DOI: https://dx.doi.org/10.2196/13216]

71. Ghafouri-Fard, S.; Mohammad-Rahimi, H.; Motie, P.; Minabi, M.A.; Taheri, M.; Nateghinia, S. Application of machine learning in the prediction of COVID-19 daily new cases: A scoping review. Heliyon; 2021; 7, e08143. [DOI: https://dx.doi.org/10.1016/j.heliyon.2021.e08143]