SarcoPred AI: data, code and models for device-free sarcopenia screening (reproducibility archive)

Published: 19 November 2026| Version 2 | DOI: 10.17632/9pc5rxv6p7.2
Contributors:
,
,
,

Description

This archive accompanies the article "SarcoPred AI, a device-free penalized logistic model for sarcopenia screening, was not outperformed by machine-learning ensembles in older adults" (Scientific Reports, under review). It contains the de-identified data of 88 community-dwelling adults aged 50 years or older assessed in Yozgat province, Türkiye, between March and May 2024 (development sample, n = 63; hold-out sample, n = 25), the complete R analysis pipeline, the final model objects and the source code of the SarcoPred AI web application (https://ai-group.shinyapps.io/sarcopredai/). Variables: age, sex, calf circumference, low 4-m gait speed (≤ 0.8 m/s), handgrip strength, a muscle mass output of the bioelectrical impedance device, and confirmed sarcopenia according to EWGSOP2 as classified by the investigators. The primary screening model uses only age, sex, calf circumference and gait speed. The variables are defined in data/data_dictionary.csv. The study was approved by the Yozgat Bozok University Social and Human Sciences Ethics Committee (decision no. 11/17, 21 February 2024), and written informed consent was obtained from all participants. Data and documentation are licensed under CC BY 4.0, and the code and application under the MIT licence (LICENSE.txt).

Files

Steps to reproduce

Install R ≥ 4.5. Set the working directory to the folder that contains data/, code/ and app/. Run source("code/00_install.R") once, then source("code/run_all.R") to reproduce every number, table and figure of the article (about 25 minutes on 30 cores; outputs are written to results/). Run source("code/verify_reproduction.R") to compare the new results with the archived results in expected_results/. To run the application locally: setwd("app"); shiny::runApp(). Details are given in README.md.

Institutions

Categories

Geriatrics, Public Health, Machine Learning, Diagnostic Testing, Sarcopenia

Licence