Risk Factors and Predictive Models for Sarcopenia in Older Adults

Abstract

ABSTRACT

Objectives

Sarcopenia as an age‐related syndrome is marked by a progressive loss of muscle strength and mass or reduced physical function. It is insidious in onset and presents a high prevalence. This study aimed to explore risk factors for sarcopenia in the elderly population and construct predictive models.

Methods

Patients (n = 335) aged 60–93 years and received an examination by a dual‐energy X‐ray absorptiometry (DXA) or a body composition analyzer (InBody) between January 2020 and May 2024 were included. Clinical data and laboratory test results were collected from these subjects. LASSO and logistic regression models were constructed to identify and evaluate significant risk factors for sarcopenia. A nomogram and a decision tree model were established for the prediction of sarcopenia probability in the elderly. Random forest was employed to rank the importance of variables in predicting sarcopenia.

Results

The potential risk factors for sarcopenia in this study were body mass index, prealbumin, albumin/globulin ratio, serum creatinine, and phosphorus. A nomogram and a decision tree model were constructed based on the factors, showing a high discriminative ability and a high classification accuracy, respectively. Both models were effective in predicting sarcopenia in the elderly, and the nomogram showed a notably reliable predictive performance.

Conclusions

This study identified risk factors and developed predictive models for sarcopenia in older adults, contributing to timely intervention and treatment of the disease. The nomogram provided an intuitive way to measure the probability of sarcopenia in the elderly population, and the decision tree model made the assessment of sarcopenia simple and rapid. Both models are helpful for clinical staff in early screening and identifying sarcopenia.

Introduction

With an aging population and increasing incidence of age-related diseases, sarcopenia has become an important health issue worldwide [1]. According to recent epidemiological data in China, the incidence of sarcopenia ranges from 5.7% to 23.9% in people aged 60 years and above [2]. Notably, sarcopenia incidence increases significantly with age [3].

Sarcopenia is considered to be a muscle disorder [4]. Its etiology is generally considered to be multifactorial, including reduced physical activity [5, 6], malnutrition [5], decreased caloric intake [7, 8], activation of inflammatory pathways [9], loss of neuromuscular junctions [10], mitochondrial abnormalities [11], reduction in satellite cell numbers [12] and hormonal changes [13]. Sarcopenia is one of the most prevalent geriatric syndromes, which is often asymptomatic and insidious in onset. It elevates the risk of mobility problems, disability, and death in the elderly [14], significantly reducing their quality of life and placing a heavy burden on family health care costs and public health expenditure [15].

Timely detection of and attention to associated risk factors are potentially critical for early detection and prevention of sarcopenia. Thus, it is vital to identify factors capable of recognizing high-risk individuals in a timely manner. This will facilitate the implementation of appropriate interventions and the improvement of prognosis. LASSO regression is a method commonly used for analysis of potential risk factors. This method can avoid the subjectivity in variable selection and address the autocorrelation and multicollinearity that are inherent in the data [16]. Additionally, it is a universal feature selection method that provides consistent performance across diverse data models [17]. Logistic regression is a well-established statistical technique that permits the analysis and modeling of binary outcomes using multiple variables. It can incorporate both categorical and continuous predictors to forecast a binary result [18]. In this study, LASSO regression and logistic regression were applied to identify risk factors for sarcopenia in older adults and develop predictive models.

Methods

Study Subjects

A comprehensive analysis was performed based on the medical records derived from elderly patients (n = 335) admitted to the Affiliated Hospital of Qingdao University between January 2020 and May 2024. Inclusion criteria: (1) age ≥ 60 years; (2) patients who provided informed consent. Exclusion criteria: (1) diseases that affect limb activity; (2) chronic consumptive diseases, such as malignancies and tuberculosis; (3) severe infectious diseases; (4) end-stage diseases; (5) communication barriers; (6) incomplete data. This study was conducted following ethics requirements.

Data Collection

Basic clinical data of the patients were collected in three parts. Demographic data, including age and sex, were collected in the first part. The second part involved blood indicators. All subjects were required to fast for overnight. The peripheral vein blood was collected in the next morning for biochemical measurements, including red blood cell (RBC) count, hemoglobin, total protein, albumin, prealbumin, globulin, albumin/globulin ratio (A/G ratio), aspartate aminotransferase/alanine aminotransferase (AST/ALT) ratio, creatinine, calcium, and phosphorus. Body mass index (BMI) and appendicular skeletal muscle mass index (ASMI) were collected in the third part. BMI is a person's weight in kilograms divided by the square of their height in metres. ASMI is a measure of muscle mass examined by either a dual-energy x-ray absorptiometry (DXA) or a body composition analyzer (InBody).

Diagnosis of Sarcopenia

The diagnosis of sarcopenia was confirmed in accordance with the Chinese Expert Consensus on Prevention and Intervention for the Elderly with Sarcopenia (2023) [2]. The diagnosis of sarcopenia was indicated based on ASMI values derived from DXA (< 7.0 kg/m² for men and < 5.4 kg/m² for women) or bioelectrical impedance analysis (BIA) (< 7.0 kg/m² for men and < 5.7 kg/m² for women).

Statistical Analysis

SPSS 26.0 and R 4.4.1 completed data analysis. Continuous variables that were normally distributed were shown as means ± standard deviations and were subject to an independent samples t-test. Measurement data with skewed distributions were described by medians and interquartile ranges (25th and 75th percentiles) and were subject to a non-parametric test. Categorical variables reported as frequencies and percentages were subject to a chi-square test. LASSO and forward stepwise regression were employed to explore significant risk factors for sarcopenia.

Lasso Regression

The R package “glmnet” (v4.4.1) was used for LASSO regression. A 10-fold cross-validation was employed to confirm the optimal penalty coefficient λ to minimize the sum of the squared errors, which balances model performance and minimizes overfitting [19]. This approach facilitated selection of optimal variables for model construction.

Model Construction and Evaluation

A training set (70%) and a validation set (30%) were generated with a random method. LASSO and logistic regression were done. A nomogram and a decision tree model were generated in the training set and validated for their predictive performance and accuracy in the validation set. The universality of each model was assessed, and a 10-fold cross-validation was used to facilitate optimal variable selection in the training set. During the construction of a decision tree model, the best split at each decision node was obtained and subjected to a test for significance. Ultimately, the probability for each subgroup of each sample was estimated.

Gini index was calculated to evaluate the importance of each variable and rank them based on the extent of improvement in the Gini index observed following their incorporation into the model [20].

Area under receiver operating characteristic (ROC-AUC) curve was calculated to assess nomogram discriminative ability. A calibration curve was plotted for evaluation of calibration. A decision curve analysis (DCA) was devised for assessment of clinical utility. The effectiveness of the decision tree model was demonstrated by a confusion matrix including accuracy, sensitivity, specificity, and other relevant metrics.

Results

Basic Characteristics

There were significant differences in BMI, RBC count, hemoglobin, albumin, prealbumin, globulin, A/G ratio, AST/ALT ratio, creatinine, phosphorus, and age between patients with and without sarcopenia (p < 0.05, Table 1).

TABLE 1 Clinical features of study subjects.

Variate	Non-sarcopenia (n = 268)	Sarcopenia (n = 67)	p
Gender
Female	161 (60.1%)	48 (71.6%)	0.080
Male	107 (39.9%)	19 (28.4%)
Age (years)	68 (64, 73)	72 (66, 78)	0.000
BMI (kg/m2)	25.20 (22.75, 27.65)	22.04 (20.10, 24.03)	0.000
RBC count (1012/L)	4.36 (4.09, 4.71)	4.15 (3.87, 4.42)	0.003
Hemoglobin (g/L)	132.00 (122.00, 141.75)	126.00 (115.00, 134.00)	0.001
Total protein (g/L)	63.78 (61.00, 67.30)	64.85 (62.10, 68.50)	0.124
Albumin (g/L)	39.99 (38.00, 41.98)	37.70 (35.40, 39.90)	0.000
Prealbumin (g/L)	254.58 ± 55.54	197.70 ± 47.67	0.000
Globulin (g/L)	24.12 (21.69, 26.79)	27.20 (24.60, 29.60)	0.000
A/G ratio	1.66 (1.49, 1.86)	1.40 (1.20, 1.54)	0.000
AST/ALT ratio	1.08 (0.83, 1.36)	1.29 (1.00, 1.70)	0.000
Creatinine (μmol/L)	54.50 (45.00, 66.00)	68.00 (54.99, 85.00)	0.000
Calcium (mmol/L)	2.28 (2.21, 2.34)	2.25 (2.17, 2.33)	0.064
Phosphorus (mmol/L)	1.15 (1.04, 1.24)	1.10 (0.95, 1.17)	0.004

LASSO Regression

LASSO regression identified variables with non-zero coefficients as potential risk factors for sarcopenia, including BMI, prealbumin, globulin, A/G ratio, creatinine, phosphorus, AST/ALT ratio, and age (Figure 1A,B). A forward stepwise multivariate logistic regression analysis was performed. BMI (OR = 0.805; 95% CI 0.731–0.887), prealbumin (OR = 0.982; 95% CI 0.975–0.990), creatinine (OR = 1.030; 95% CI 1.014–1.047), A/G ratio (OR = 0.146; 95% CI 0.038–0.563) and phosphorus (OR = 0.101; 95% CI 0.012–0.854) were identified as independent risk factors for sarcopenia (p < 0.05, Table 2).

[IMAGE OMITTED. SEE PDF]

TABLE 2 Logistic regression analysis of risk factors for sarcopenia.

Variate	OR	95% CI	p
BMI	0.805	0.731, 0.887	< 0.001
Prealbumin	0.982	0.975, 0.990	< 0.001
Creatinine	1.030	1.014, 1.047	< 0.001
A/G ratio	0.146	0.038, 0.563	0.005
Phosphorus	0.101	0.012, 0.854	0.035

Nomogram Construction and Evaluation

A nomogram was constructed incorporating the five independent risk factors using the R package “rms” (v4.4.1). The model achieved an optimal score of 350. A score above 230 indicates an 80% probability of sarcopenia (Figure 2).

[IMAGE OMITTED. SEE PDF]

The AUC was calculated as 0.896 (95% CI: 0.841–0.950) in the training set and 0.848 (95% CI: 0.760–0.936) in the validation set (Figure 3A,D). Calibration analysis revealed a high degree of concordance between observed outcomes and predicted probabilities in both sets (Figure 3B,E). DCA analysis of the nomogram suggested favorable net benefit (Figure 3C,F).

[IMAGE OMITTED. SEE PDF]

Decision Tree Model Construction

A decision tree model based on 5 significant risk factors was generated in the training set. A complexity parameter (cp) value of 0.02 was used to prune the decision tree. A/G ratio, BMI, creatinine, and prealbumin were identified as key factors discriminative for sarcopenia. A decision threshold of 0.2 was applied. The classification ability of the model was evaluated in the validation set, resulting in a classification accuracy of 78%, sensitivity of 58.82%, and specificity of 81.93% (Figure 4).

[IMAGE OMITTED. SEE PDF]

Random Forest Model Construction

A random forest model was developed using an ensemble of 500 decision trees. The model identified the nine most significant predictors of sarcopenia as BMI, prealbumin, creatinine, A/G ratio, globulin, albumin, age, AST/ALT ratio, and phosphorus (Figure 5).

[IMAGE OMITTED. SEE PDF]

Discussion

Sarcopenia is a common geriatric syndrome marked by a gradual loss of muscle mass and strength. It is a common cause of muscle wasting, reduced mobility, fatigue, and difficulty walking in older adults. Because of the lack of obvious and specific clinical symptoms in the early stages, early detection of sarcopenia is not easy. Thus, timely detection of and attention to associated risk factors are potentially critical for early detection and prevention of sarcopenia. This study explored risk factors for sarcopenia in the elderly population and developed predictive models with clinical significance.

This study applied multiple models, including LASSO regression, logistic regression, a nomogram, a decision tree model, and a random forest model. LASSO regression was applied to resolve multicollinearity and obtain potential factors predicting sarcopenia [19]. However, it may exclude weak but significant variables and is highly sensitive to the choice of penalty parameter λ [21]. A logistic regression model was established to obtain independent risk factors for sarcopenia (BMI, prealbumin, creatinine, A/G ratio, and phosphorus), but the linear assumption required in logistic regression limits its application in datasets with complex, nonlinear interactions [18]. A nomogram based on independent risk factors for sarcopenia was established with high precision and superior predictive performance. The AUC was 0.896 in the training set and 0.848 in the validation set. Despite its advantages, the use of a nomogram requires high-quality data and has a steep learning curve for clinicians. Decision tree models provide simplicity and a clear process of decision-making, but they are prone to overfitting and may have low accuracy in heterogeneous datasets [22, 23]. A random forest model was constructed using an ensemble of 500 decision trees. This model tolerates highly correlated predictors and models nonlinear relationships and higher-order interactions without the need for predefined functional forms [24]. However, its black-box nature limits interpretability, and there are no explicit equations linking variables to risk estimates. In addition, the complexity of constructing random forest models and the variability of models across datasets make parameter comparison and validation in independent population cohorts difficult [24].

In this study, a LASSO regression model was generated to obtain the optimal set of risk factors for sarcopenia. Variables exhibiting collinearity and weak predictive strength were excluded. Consequently, BMI, prealbumin, globulin, A/G ratio, creatinine, phosphorus, AST/ALT ratio, and age were selected as potential risk factors. A forward stepwise logistic regression identified BMI, prealbumin, creatinine, phosphorus, and A/G ratio as independent risk factors for sarcopenia. Based on these factors, a nomogram model and a decision tree model were constructed. ROC analysis and assessment of classification accuracy demonstrated a high performance of the nomogram and the decision tree model in predicting sarcopenia in the elderly. Notably, the nomogram exhibited superior predictive capabilities, while the decision tree model was better in terms of simplicity of use and rapid implementation in routine clinical practice for the assessment of sarcopenia.

This study showed that a lower BMI predicted an increased risk of sarcopenia, in accordance with previous findings. It has been suggested that a BMI of 21 kg/m² may be a cut-off point for predicting sarcopenia in the elderly [25–27]. Nevertheless, a higher value of BMI is not always beneficial. While it may reduce the risk of sarcopenia, it may also increase the risk of metabolic syndrome. Testosterone plays a notable role in the metabolic syndrome. It affects muscle composition, fat distribution and insulin sensitivity [28]. Obesity-related metabolic disorders can exacerbate malnutrition, creating a vicious circle.

This study also revealed that prealbumin and A/G ratio were risk factors for sarcopenia. Sarcopenia is complex in etiology, and malnutrition is considered to be a major contributing factor [13]. Albumin and prealbumin are serum proteins and common biomarkers used to assess nutritional status in individuals [29, 30]. Levels of either or both below the normal range may indicate malnutrition [31]. Prealbumin is a small-molecular-weight protein synthesized by the liver and able to rapidly reflect changes in nutritional status [16]. Previous research revealed that the albumin level was positively linked to walking speed and grip strength, and a lower albumin level suggested a decreased muscle strength [32, 33].

Inflammation and immunosenescence are prevalent features of ageing. Albumin has specific clinical implications in inflammatory conditions [33, 34]. Globulins are composed of a number of proteins involved in the inflammatory process and show an increasing trend in inflammatory conditions [35]. A/G ratio is a more accurate indicator of nutritional and inflammatory status [36, 37]. Chen et al. [38] found that people with reduced muscle mass exhibited a notable increase in globulin and a marked decrease in albumin and A/G ratio. Additionally, albumin and A/G ratio were positively linked to muscle mass in men, while globulin revealed a negative correlation. Future research is required to establish the causal relationship between albumin, globulin, A/G ratio, and muscle mass, and identify the underlying mechanism.

Serum creatinine was also confirmed as a factor predicting sarcopenia. Serum creatinine is the metabolite of phosphocreatine in skeletal muscle that correlates to the amount of muscle tissue [39]. Serum creatinine and cystatin C have been employed in the calculation of the Sarcopenia Index (SI), which provides an approximation of skeletal muscle mass and may help assess sarcopenia [40, 41]. Previous research showed a potential link between sarcopenia and muscle function, and a positive linkage between sarcopenia and handgrip strength and gait speed was demonstrated [42]. Another study found that SI was unable to accurately detect skeletal muscle mass or sarcopenia defined by bioelectrical impedance analysis (BIA), particularly in older adults with normal renal function [43]. Thus, despite the potential linkage to muscle mass and function, creatinine may have limited accuracy in independently diagnosing sarcopenia.

Blood phosphorus was another risk factor for sarcopenia identified in this study. Hypophosphatemia is associated with muscle damage, and muscle weakness is one of the most prominent symptoms of hypophosphatemia [44, 45]. It is well known that vitamin D deficiency can cause a decline in muscle strength [46], and that simultaneous deficiency and hypophosphatemia can significantly decrease muscle strength [47].

In all, this study explored potential risk factors for sarcopenia in older adults using LASSO and logistic regression models. Meanwhile, a nomogram and a decision tree model were generated to predict the probability of sarcopenia. This study provides insights into the early detection of sarcopenia, but there are some limitations. Further studies to elucidate the causal sequence between risk factors and the onset of sarcopenia are important for understanding the etiology of sarcopenia. Imaging modalities (e.g., computed tomography or magnetic resonance imaging) can be incorporated into the predictive models in future research to further improve accuracy and provide new ways to assess muscle mass and monitor disease progression.

Conclusion

To conclude, BMI, prealbumin, A/G ratio, serum creatinine, and phosphorus were independent predictors of sarcopenia. The nomogram and decision tree model based on these factors may help clinicians identify sarcopenia early and facilitate timely intervention and treatment.

Author Contributions

Wenjing Feng, Weiguo Chen, and Yongjun Mao contributed to the conception of the study. Shiyuan Zhang, Nina An, Meng Lv, and Lanyu Yang facilitated the data collection and data handling. Shiyuan Zhang, Xue Yang, Rui Liu, and Song Hu performed the data analyses and interpretation. Wenjing Feng and Weiguo Chen reviewed and revised the manuscript. Shiyuan Zhang and Xue Yang contributed significantly to writing the manuscript. All authors contributed to the article and approved the submitted version.

Acknowledgments

The authors have nothing to report.

Ethics Statement

The Medical Ethics Committee of the Affiliated Hospital of Qingdao University (reference number QYFYWZLL29628) approved the study protocol.

Conflicts of Interest

The authors declare no conflicts of interest.

References

A. J. Cruz‐Jentoft and A. A. Sayer, “Sarcopenia,” Lancet 393, no. 10191 (2019): 2636–2646, https://doi.org/10.1016/S0140‐6736(19)31138‐9.

H. Cui, Z. Wang, J. Wu, et al., “Chinese Expert Consensus on Prevention and Intervention for Elderly With Sarcopenia (2023),” Aging Medicine 6, no. 2 (2023): 104–115, https://doi.org/10.1002/agm2.12245.

P. Han, J. Zhao, Q. Guo, et al., “Incidence, Risk Factors, and the Protective Effect of High Body Mass Index Against Sarcopenia in Suburb‐Dwelling Elderly Chinese Populations,” Journal of Nutrition, Health & Aging 20, no. 10 (2016): 1056–1060, https://doi.org/10.1007/s12603‐016‐0704‐3.

A. J. Cruz‐Jentoft, G. Bahat, J. Bauer, et al., “Sarcopenia: Revised European Consensus on Definition and Diagnosis,” Age and Ageing 48, no. 1 (2019): 16–31, https://doi.org/10.1093/ageing/afy169.

G. C. Ligthart‐Melis, Y. C. Luiking, A. Kakourou, T. Cederholm, A. B. Maier, and M. A. E. de van r Schueren, “Frailty, Sarcopenia, and Malnutrition Frequently (Co‐)occur in Hospitalized Older Adults: A Systematic Review and Meta‐Analysis,” Journal of the American Medical Directors Association 21, no. 9 (2020): 1216–1228, https://doi.org/10.1016/j.jamda.2020.03.006.

Y. Shen, Q. Shi, K. Nong, et al., “Exercise for Sarcopenia in Older People: A Systematic Review and Network Meta‐Analysis,” Journal of Cachexia, Sarcopenia and Muscle 14, no. 3 (2023): 1199–1211, https://doi.org/10.1002/jcsm.13225.

E. Marty, Y. Liu, A. Samuel, O. Or, and J. Lane, “A Review of Sarcopenia: Enhancing Awareness of an Increasingly Prevalent Disease,” Bone 105 (2017): 276–286, https://doi.org/10.1016/j.bone.2017.09.008.

W. F. Nieuwenhuizen, H. Weenen, P. Rigby, and M. M. Hetherington, “Older Adults and Patients in Need of Nutritional Support: Review of Current Treatment Options and Factors Influencing Nutritional Intake,” Clinical Nutrition 29, no. 2 (2010): 160–169, https://doi.org/10.1016/j.clnu.2009.09.003.

L. A. Schaap, S. M. F. Pluijm, D. J. H. Deeg, et al., “Higher Inflammatory Marker Levels in Older Persons: Associations With 5‐Year Change in Muscle Mass and Muscle Strength,” Journals of Gerontology. Series A, Biological Sciences and Medical Sciences 64, no. 11 (2009): 1183–1189, https://doi.org/10.1093/gerona/glp097.

L. B. Verdijk, M. L. Dirks, T. Snijders, et al., “Reduced Satellite Cell Numbers With Spinal Cord Injury and Aging in Humans,” Medicine and Science in Sports and Exercise 44, no. 12 (2012): 2322–2330, https://doi.org/10.1249/MSS.0b013e3182667c2e.

E. Ferri, E. Marzetti, R. Calvani, A. Picca, M. Cesari, and B. Arosio, “Role of Age‐Related Mitochondrial Dysfunction in Sarcopenia,” International Journal of Molecular Sciences 21, no. 15 (2020): 5236, https://doi.org/10.3390/ijms21155236.

L. González‐Blanco, M. Bermúdez, J. C. Bermejo‐Millo, et al., “Cell Interactome in Sarcopenia During Aging,” Journal of Cachexia, Sarcopenia and Muscle 13, no. 2 (2022): 919–931, https://doi.org/10.1002/jcsm.12937.

J. D. Walston, “Sarcopenia in Older Adults,” Current Opinion in Rheumatology 24, no. 6 (2012): 623–627.

S. S. Y. Yeung, E. M. Reijnierse, V. K. Pham, et al., “Sarcopenia and Its Association With Falls and Fractures in Older Adults: A Systematic Review and Meta‐Analysis,” Journal of Cachexia, Sarcopenia and Muscle 10, no. 3 (2019): 485–500, https://doi.org/10.1002/jcsm.12411.

Q. Li, H. Cheng, W. Cen, T. Yang, and S. Tao, “Development and Validation of a Predictive Model for the Risk of Sarcopenia in the Older Adults in China,” European Journal of Medical Research 29, no. 1 (2024): 278, https://doi.org/10.1186/s40001‐024‐01873‐w.

J. K. Tay, B. Narasimhan, and T. Hastie, “Elastic Net Regularization Paths for all Generalized Linear Models,” Journal of Statistical Software 106 (2023): 1, https://doi.org/10.18637/jss.v106.i01.

S. Mirjalili, P. Powell, J. Strunk, T. James, and A. Duarte, “Evaluation of Classification Approaches for Distinguishing Brain States Predictive of Episodic Memory Performance From Electroencephalography: Abbreviated Title: Evaluating Methods of Classifying Memory States From EEG,” NeuroImage 247 (2022): 118851, https://doi.org/10.1016/j.neuroimage.2021.118851.

M. E. Shipe, S. A. Deppen, F. Farjah, and E. L. Grogan, “Developing Prediction Models for Clinical Use Using Logistic Regression: An Overview,” Journal of Thoracic Disease 11, no. Suppl 4 (2019): S574–S584, https://doi.org/10.21037/jtd.2019.01.25.

A. J. McEligot, V. Poynor, R. Sharma, and A. Panangadan, “Logistic LASSO Regression for Dietary Intakes and Breast Cancer,” Nutrients 12, no. 9 (2020): 2652, https://doi.org/10.3390/nu12092652.

B. H. Menze, B. M. Kelm, R. Masuch, et al., “A Comparison of Random Forest and Its Gini Importance With Standard Chemometric Methods for the Feature Selection and Classification of Spectral Data,” BMC Bioinformatics 10 (2009): 213, https://doi.org/10.1186/1471‐2105‐10‐213.

C. Zou, L. Su, M. Pan, et al., “Exploration of Novel Biomarkers in Alzheimer's Disease Based on Four Diagnostic Models,” Frontiers in Aging Neuroscience 15 (2023): 1079433, https://doi.org/10.3389/fnagi.2023.1079433.

M. P. Romero, Y. M. Chang, L. A. Brunton, et al., “Decision Tree Machine Learning Applied to Bovine Tuberculosis Risk Factors to Aid Disease Control Decision Making,” Preventive Veterinary Medicine 175 (2020): 104860, https://doi.org/10.1016/j.prevetmed.2019.104860.

B. Osong, I. Bermejo, K. C. Lee, S. H. Lee, A. Dekker, and J. van Soest, “Prediction of Radiotherapy Compliance in Elderly Cancer Patients Using an Internally Validated Decision Tree,” Cancers (Basel) 14, no. 24 (2022): 6116, https://doi.org/10.3390/cancers14246116.

L. Langsetmo, J. T. Schousboe, B. C. Taylor, et al., “Advantages and Disadvantages of Random Forest Models for Prediction of Hip Fracture Risk Versus Mortality Risk in the Oldest Old,” JBMR Plus 7, no. 8 (2023): e10757, https://doi.org/10.1002/jbm4.10757.

L. C. Wu, H. H. Kao, H. J. Chen, and P. F. Huang, “Preliminary Screening for Sarcopenia and Related Risk Factors Among the Elderly,” Medicine (Baltimore) 100, no. 19 (2021): e25946, https://doi.org/10.1097/MD.0000000000025946.

U. G. Kyle, M. Pirlich, T. Schuetz, H. J. Luebke, H. Lochs, and C. Pichard, “Prevalence of Malnutrition in 1760 Patients at Hospital Admission: A Controlled Population Study of Body Composition,” Clinical Nutrition 22, no. 5 (2003): 473–481, https://doi.org/10.1016/S0261‐5614(03)00049‐9.

H. Kim, “Incidence and Predictors of Sarcopenia Onset in Community‐Dwelling Elderly Japanese Women: 4‐Year Follow‐Up Study,” Heart Disease 16 (2015): 85.

V. Xega, and J. L. Liu, “Beyond Reproduction: Unraveling the Impact of Sex Hormones on Cardiometabolic Health,” Medical Review 4, no. 4 (2024): 284–300, https://doi.org/10.1515/mr‐2024‐0012.

Z. Zhang, S. L. Pereira, M. Luo, and E. M. Matheson, “Evaluation of Blood Biomarkers Associated With Risk of Malnutrition in Older Adults: A Systematic Review and Meta‐Analysis,” Nutrients 9 (2017): 8, https://doi.org/10.3390/nu9080829.

G. Sergi, A. Coin, G. Enzi, et al., “Role of Visceral Proteins in Detecting Malnutrition in the Elderly,” European Journal of Clinical Nutrition 60, no. 2 (2006): 203–209, https://doi.org/10.1038/sj.ejcn.1602289.

G. Devoto, F. Gallo, C. Marchello, et al., “Prealbumin Serum Concentrations as a Useful Tool in the Assessment of Malnutrition in Hospitalized Patients,” Clinical Chemistry 52, no. 12 (2006): 2281–2285, https://doi.org/10.1373/clinchem.2006.080366.

C. K. Snyder, J. A. Lapidus, P. M. Cawthon, T. T. L. Dam, L. Y. Sakai, and L. M. Marshall, “Serum Albumin in Relation to Change in Muscle Mass, Muscle Strength, and Muscle Power in Older Men,” Journal of the American Geriatrics Society 60, no. 9 (2012): 1663–1672, https://doi.org/10.1111/j.1532‐5415.2012.04115.x.

V. A. van Atteveld, J. M. van Ancum, E. M. Reijnierse, M. C. Trappenburg, C. G. M. Meskers, and A. B. Maier, “Erythrocyte Sedimentation Rate and Albumin as Markers of Inflammation Are Associated With Measures of Sarcopenia: A Cross‐Sectional Study,” BMC Geriatrics 19, no. 1 (2019): 233, https://doi.org/10.1186/s12877‐019‐1253‐5.

L. H. Jin, L. S. Shi, and W. D. Huang, “The Role of Bile Acids in Human Aging,” Medical Review 4, no. 2 (2024): 154–157, http://doi.org/10.1515/mr‐2024‐0003.

T. Zhou, S. T. Yu, W. Z. Chen, R. Xie, and J. C. Yu, “Pretreatment Albumin Globulin Ratio has a Superior Prognostic Value in Laryngeal Squamous Cell Carcinoma Patients: A Comparison Study,” Journal of Cancer 10, no. 3 (2019): 594–601, https://doi.org/10.7150/jca.28817.

P. P. Wu, Y. P. Hsieh, C. T. Kor, and P. F. Chiu, “Association Between Albumin‐Globulin Ratio and Mortality in Patients With Chronic Kidney Disease,” Journal of Clinical Medicine 8, no. 11 (2019): 1991, https://doi.org/10.3390/jcm8111991.

S. Maeda, Y. Takeya, R. Oguro, et al., “Serum Albumin/Globulin Ratio Is Associated With Cognitive Function in Community‐Dwelling Older People: The Septuagenarians, Octogenarians, Nonagenarians Investigation With Centenarians Study,” Geriatrics & Gerontology International 19, no. 10 (2019): 967–971, https://doi.org/10.1111/ggi.13751.

Z. Chen, C. Song, Z. Yao, J. Sun, and W. Liu, “Associations Between Albumin, Globulin, Albumin to Globulin Ratio and Muscle Mass in Adults: Results From the National Health and Nutrition Examination Survey 2011–2014,” BMC Geriatrics 22, no. 1 (2022): 383, https://doi.org/10.1186/s12877‐022‐03094‐4.

M. Tosato, E. Marzetti, M. Cesari, et al., “Measurement of Muscle Mass in Sarcopenia: From Imaging to Biochemical Markers,” Aging Clinical and Experimental Research 29, no. 1 (2017): 19–27, https://doi.org/10.1007/s40520‐016‐0717‐0.

K. B. Kashani, E. N. Frazee, L. Kukrálová, et al., “Evaluating Muscle Mass by Using Markers of Kidney Function: Development of the Sarcopenia Index,” Critical Care Medicine 45, no. 1 (2017): e23–e29, https://doi.org/10.1097/CCM.0000000000002013.

Y. L. Lin, S. Y. Chen, Y. H. Lai, et al., “Serum Creatinine to Cystatin C Ratio Predicts Skeletal Muscle Mass and Strength in Patients With Non‐Dialysis Chronic Kidney Disease,” Clinical Nutrition 39, no. 8 (2020): 2435–2441, https://doi.org/10.1016/j.clnu.2019.10.027.

T. Tang, Y. Zhuo, L. Xie, H. Wang, and M. Yang, “Sarcopenia Index Based on Serum Creatinine and Cystatin C Is Associated With 3‐Year Mortality in Hospitalized Older Patients,” Scientific Reports 10, no. 1 (2020): 1260, https://doi.org/10.1038/s41598‐020‐58304‐z.

Q. He, J. Jiang, L. Xie, L. Zhang, and M. Yang, “A Sarcopenia Index Based on Serum Creatinine and Cystatin C Cannot Accurately Detect Either Low Muscle Mass or Sarcopenia in Urban Community‐Dwelling Older People,” Scientific Reports 8, no. 1 (2018): 11534, https://doi.org/10.1038/s41598‐018‐29808‐6.

Y. Y. Chen, T. W. Kao, C. W. Chou, et al., “Exploring the Link Between Serum Phosphate Levels and Low Muscle Strength, Dynapenia, and Sarcopenia,” Scientific Reports 8, no. 1 (2018): 3573, https://doi.org/10.1038/s41598‐018‐21784‐1.

D. H. Pesta, D. N. Tsirigotis, D. E. Befroy, et al., “Hypophosphatemia Promotes Lower Rates of Muscle ATP Synthesis,” FASEB Journal 30, no. 10 (2016): 3378–3387, https://doi.org/10.1096/fj.201600473R.

M. Kupisz‐Urbańska, P. Płudowski, and E. Marcinowska‐Suchowierska, “Vitamin D Deficiency in Older Patients‐Problems of Sarcopenia, Drug Interactions, Management in Deficiency,” Nutrients 13, no. 4 (2021): 1247, https://doi.org/10.3390/nu13041247.

L. Schubert and H. F. DeLuca, “Hypophosphatemia Is Responsible for Skeletal Muscle Weakness of Vitamin D Deficiency,” Archives of Biochemistry and Biophysics 500, no. 2 (2010): 157–161, https://doi.org/10.1016/j.abb.2010.05.029.