|
ИСТИНА |
Войти в систему Регистрация |
Интеллектуальная Система Тематического Исследования НАукометрических данных |
||
Machine Learning-Based Decision-Making in Geriatrics: Aging Phenotype Calculator and Survival Prognosisстатья Исследовательская статья
- Авторы: Mamchur Aleksandra, Sharashkina Natalia, Erema Veronika, Kashtanova Daria, Ivanov Mikhail, Bruttan Maria, Zelenova Elena, Shelly Eva, Ostapenko Valentina, Dzhumaniiazova Irina, Matkava Lorena, Yudin Vladimir, Akopyan Anna, Strazhesko Irina, Maytesyan Lilit, Tarasova Irina, Beloshevskaya Olga, Keskinov Anton, Tkacheva Olga, Yudin Sergey
- Журнал: Aging and disease
- Том: 16
- Номер: 1
- Год издания: 2025
- Издательство: JKL International
- Местоположение издательства: California, United States
- DOI: 10.14336/AD.2024.0120
- Аннотация: Aging is a natural process with varying effects. As we grow older, our bodies become more susceptible to aging-associated diseases. These diseases, individually or collectively, lead to the formation of distinct aging phenotypes. Identifying these aging phenotypes and understanding the complex interplay between coexistent diseases would facilitate more personalized patient management, a better prognosis, and a prolonged lifespan. Many studies distinguish between successful aging and frailty. However, this simple distinction fails to reflect the diversity of underlying causes. In this study, we sought to establish the underlying causes of frailty and determine the patterns in which these causes converge to form aging phenotypes. We conducted a comprehensive geriatric examination, cognitive assessment, and survival analysis of 2,688 long-living adults (median age = 92 years). The obtained data were clustered and used as input data for the Aging Phenotype Calculator, a multiclass classification model validated on an independent dataset of 96 older adults. The accuracy of the model was assessed using the receiver operating characteristic curve and the area under the curve. Additionally, we analyzed socioeconomic factors that could contribute to specific aging patterns. We identified five aging phenotypes: non-frailty, multimorbid frailty, metabolic frailty, cognitive frailty, and functional frailty. For each phenotype, we determined the underlying diseases and conditions and assessed the survival rate. Additionally, we provided management recommendations for each of the five phenotypes based on their distinct features and associated challenges. The identified aging phenotypes may facilitate better-informed decision-making. The Aging Phenotype Calculator (ROC AUC = 92%) may greatly assist geriatricians in patient management.
- Добавил в систему: Мамчур Александра Александровна