YUE Yibing, YU Ying, SHEN Lei, WANG Yan, WANG Yingying, LYU Weibo, LIU Chuang. Prediction of sarcopenia based on longitudinal physical examination data[J]. CHINESE JOURNAL OF DISEASE CONTROL & PREVENTION, 2023, 27(11): 1296-1299. doi: 10.16462/j.cnki.zhjbkz.2023.11.009
Citation:
YUE Yibing, YU Ying, SHEN Lei, WANG Yan, WANG Yingying, LYU Weibo, LIU Chuang. Prediction of sarcopenia based on longitudinal physical examination data[J]. CHINESE JOURNAL OF DISEASE CONTROL & PREVENTION, 2023, 27(11): 1296-1299. doi: 10.16462/j.cnki.zhjbkz.2023.11.009
YUE Yibing, YU Ying, SHEN Lei, WANG Yan, WANG Yingying, LYU Weibo, LIU Chuang. Prediction of sarcopenia based on longitudinal physical examination data[J]. CHINESE JOURNAL OF DISEASE CONTROL & PREVENTION, 2023, 27(11): 1296-1299. doi: 10.16462/j.cnki.zhjbkz.2023.11.009
Citation:
YUE Yibing, YU Ying, SHEN Lei, WANG Yan, WANG Yingying, LYU Weibo, LIU Chuang. Prediction of sarcopenia based on longitudinal physical examination data[J]. CHINESE JOURNAL OF DISEASE CONTROL & PREVENTION, 2023, 27(11): 1296-1299. doi: 10.16462/j.cnki.zhjbkz.2023.11.009
Objective In the elderly, sarcopenia is a common disease and efficient identification of sarcopenia is very important to keep health.Methods Based on the longitudinal physical examination data, a total of 2 544 subjects in a hospital from Shanghai during 2013-2019 were included. Considering the difference across annual indices, various machine learning models were constructed to predict the risk of sarcopenia in the elderly, and the decision curve analysis was applied to provide reference for clinical decision makers.Results The prediction models results showed that the prediction accuracy based on the Light Gradient Boosting Machine (LightGBM) model was relatively high (AUC=0.913 4). Decision curve analysis indicated that the net profit of the LightGBM model became larger when threshold probability (threshold for judging sarcopenia) ranged from 0.01 to 0.42 and from 0.84 to 0.92. And the net profit of the Random Forest model was larger when threshold probability ranged from 0.42 to 0.50 and from 0.60 to 0.67, while in the case of logistic regression model, the range was located in 0.50-0.60 and 0.67-0.84.Conclusions The prediction model established based on longitudinal physical examination data and machine learning methods can effectively predict the future risk of sarcopenia in the elderly, and is of great value for the early diagnosis and intervention of sarcopenia.
Wiedmer P, Jung T, Castro JP, et al. Sarcopenia- Molecular mechanisms and open questions[J]. Ageing Res Rev, 2021, 65: 101200. DOI: 10.1016/j.arr.2020.101200.
[2]
Beaudart C, Zaaria M, Pasleau F, et al. Health outcomes of sarcopenia: a systematic review and meta-analysis[J]. PLoS One, 2017, 12(1): e0169548. DOI: 10.1371/journal.pone.0169548.
[3]
Bruyère O, Beaudart C, Ethgen O, et al. The health economics burden of sarcopenia: a systematic review[J]. Maturitas, 2019, 119: 61-69. DOI: 10.1016/j.maturitas.2018.11.003.
[4]
Rajkomar A, Dean J, Kohane I. Machine learning in medicine[J]. N Engl J Med, 2019, 380(14): 1347-1358. DOI: 10.1056/NEJMra1814259.
Ou YP, Li XX, Leng F, et al. Application of machi-ne learning algorithm in diabetes risk prediction of physical examination population[J]. Chin J Dis Control Prev, 2021, 25(7): 849-853, 868. DOI: 10.16462/j.cnki.zhjbkz.2021.07.020.
Liu ZS, Siu WC, Chan YL. Features guided face super-resolution via hybrid model of deep learning and random forests[J]. IEEE Trans Image Process, 2021, 30: 4157-4170. DOI: 10.1109/TIP.2021.3069554.
[8]
Ke GL, Meng Q, Finley T, et al. LightGBM: a highly efficient gradient boosting decision tree[J]. NeurIPS Curran Associates Inc, 2017: 3149-3157.
[9]
Vickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models[J]. Med Decis Making, 2006, 26(6): 565-574. DOI: 10.1177/0272989X06295361.