Machine Learning Approaches for Predicting Frailty Base on Multimorbidities in US Adults Using NHANES Data (1999-2018)
Computer Methods and Programs in Biomedicine Update
Cite this paper
Plain citation
Teng Li, Xueke Li, Haoran XU, Yanyan Wang, Jingyu Ren, Shixiang Jing, Zichen Jin, Gang chen, Youyou Zhai, Zeyu Wu, Ge Zhang, Yuying Wang. Machine Learning Approaches for Predicting Frailty Base on Multimorbidities in US Adults Using NHANES Data (1999-2018). Computer Methods and Programs in Biomedicine Update. 2024. https://doi.org/10.1016/j.cmpbup.2024.100164
Research context and source limits
Limited evidence note. Only the source material identified below has been reviewed; this is not a complete Paper GEO V2 assessment.
Cross-sectional survey methods and frailty-label definition; sample denominator and model validation remain unresolved.
Research question and data source
This study examines multimorbidity and frailty using NHANES survey cycles from 1999–2018. The paper describes cross-sectional analyses and machine-learning classification, rather than prospective prediction of new frailty or evaluation of a preventive treatment. Its scientific question concerns associations and classification within the analysed survey records.
Source: Methods 2: data source and participant selection; Methods 2.3
Frailty definition and analysis
Frailty is defined through a modified 36-item deficit-accumulation index. The Methods describe survey-weighted summaries, regression analyses and resampling for model selection. The complete item list and predictor overlap still require verification; reporting these procedures does not establish that outcome leakage was excluded or that an independent external cohort was tested.
Source: Methods 2.2: Assessment of disease and frailty; Methods 2.3; Supplementary materials
Unresolved analytical denominator
Methods reports 42,886 participants, while Results states 46,187 and then refers to 431 participants. Table 1 sex counts support a different total from the Results headline. These statements cannot be reconciled by selecting the most plausible number. The model-evaluation partition also remains insufficiently specified.
Source: Methods 2 first paragraph; Results 3 first paragraph; Table 1 Sex rows; Results 3.2
Evidence limits
- No participant total or predictive-performance estimate is endorsed.
- Corrected flow, outcome items and resampling details are needed for complete V2.
Provenance
- Article source — Official publisher or PMC full text reviewed locally. Methods 2: data source and participant selection; Methods 2.3
Source review: 2026-10-27.
Verify publication identity
These records identify the same original work; database inclusion is not evidence of scientific validity or clinical benefit.