A Multivariable Prediction Model for Mild Cognitive Impairment and Dementia: Algorithm Development and Validation (Preprint)

Published: April 11, 2024

Mild cognitive impairment (MCI) poses significant challenges in early diagnosis and timely intervention. Underdiagnosis, coupled with the economic social burden of dementia, necessitates more precise detection methods. Machine learning (ML) algorithms show promise managing complex data for MCI dementia prediction. This study assessed predictive accuracy ML models identifying onset using Korean Longitudinal Study Aging (KLoSA) dataset. used from KLoSA, a comprehensive biennial survey that tracks demographic, health, socioeconomic aspects middle-aged older adults 2018 to 2020. Among 6171 initial households, 4975 eligible adult participants aged 60 years or were selected after excluding individuals based on age missing data. The identification relied self-reported diagnoses, sociodemographic health-related variables serving as key covariates. dataset was categorized into training test sets predict by multiple models, including logistic regression, light gradient-boosting machine, XGBoost (extreme gradient boosting), CatBoost, random forest, boosting, AdaBoost, support vector classifier, k-nearest neighbors, evaluate performance. performance area under receiver operating characteristic curve (AUC). Class imbalances addressed via weights. Shapley additive explanation values determine contribution each feature prediction rate. participants, best model predicting median AUC 0.6729 (IQR 0.3883-0.8152), followed neighbors 0.5576 0.4555-0.6761) classifier 0.5067 0.3755-0.6389). For prediction, XGBoost, achieving 0.8185 0.8085-0.8285), closely machine 0.8069 0.7969-0.8169) AdaBoost 0.8007 0.7907-0.8107). highlighted pain everyday life, being widowed, living alone, exercising, partner strongest predictors MCI. most features other contributing factors, education at high school level, middle monthly engagement. algorithms, especially exhibited potential KLoSA However, no has demonstrated robust dementia. Sociodemographic factors are crucial initiating conditions, emphasizing need multifaceted These findings underscore limitations community-dwelling adults.

Language: Английский

A Multivariable Prediction Model for Mild Cognitive Impairment and Dementia: Algorithm Development and Validation

JMIR Medical Informatics, Journal Year: 2024, Volume and Issue: 12, P. e59396 - e59396

Published: Nov. 22, 2024

Background Mild cognitive impairment (MCI) poses significant challenges in early diagnosis and timely intervention. Underdiagnosis, coupled with the economic social burden of dementia, necessitates more precise detection methods. Machine learning (ML) algorithms show promise managing complex data for MCI dementia prediction. Objective This study assessed predictive accuracy ML models identifying onset using Korean Longitudinal Study Aging (KLoSA) dataset. Methods used from KLoSA, a comprehensive biennial survey that tracks demographic, health, socioeconomic aspects middle-aged older adults 2018 to 2020. Among 6171 initial households, 4975 eligible adult participants aged 60 years or were selected after excluding individuals based on age missing data. The identification relied self-reported diagnoses, sociodemographic health-related variables serving as key covariates. dataset was categorized into training test sets predict by multiple models, including logistic regression, light gradient-boosting machine, XGBoost (extreme gradient boosting), CatBoost, random forest, boosting, AdaBoost, support vector classifier, k-nearest neighbors, evaluate performance. performance area under receiver operating characteristic curve (AUC). Class imbalances addressed via weights. Shapley additive explanation values determine contribution each feature prediction rate. Results participants, best model predicting median AUC 0.6729 (IQR 0.3883-0.8152), followed neighbors 0.5576 0.4555-0.6761) classifier 0.5067 0.3755-0.6389). For prediction, XGBoost, achieving 0.8185 0.8085-0.8285), closely machine 0.8069 0.7969-0.8169) AdaBoost 0.8007 0.7907-0.8107). highlighted pain everyday life, being widowed, living alone, exercising, partner strongest predictors MCI. most features other contributing factors, education at high school level, middle monthly engagement. Conclusions algorithms, especially exhibited potential KLoSA However, no has demonstrated robust dementia. Sociodemographic factors are crucial initiating conditions, emphasizing need multifaceted These findings underscore limitations community-dwelling adults.

Language: Английский