|
KMID : 1161020230050020065
|
|
Korea Journal of Geriatric Occupational Therapy
2023 Volume.5 No. 2 p.65 ~ p.83
|
|
|
Exploring Lifestyle Factors Affecting Frailty Among Older Adults: Using a Decision Tree and Tandom Forest Methodology
|
|
Ham Yea-Jin
Park Ji-Hyuk
|
|
|
Abstract
|
|
|
Objective : This study aims to identify the risk and preventive lifestyle factors in which healthy older adults can affect frailty by using machine learning ― notably, decision trees and random forests.
Methods : We used the 2020 Korean Longitudinal Study of Aging data to identify lifestyle factors affecting older adults' frailty, applying decision trees and random forest analysis. Four variables were used to distinguish the criteria for frailty, along with 26 variables related to lifestyle factors ― including demographic information.
Results : Of a total of 26 variables in the decision tree, four variables (Mini-mental State Examination scores, frequency of meeting with people, sex, and level of depression) were selected as the model. In the random forest analysis, “frequency of meeting with people” the “Mini-mental State Examination score,” the “Geriatric Oral Health Assessment Index,” “subjective health status,” “age,” “sex,” “the number of travel/tourism/outdoor experiences,” and “education level” were selected as important variables. In the partial dependence analysis, pre-frailty displayed a transitional pattern of non-frailty, turning into frailty.
Conclusion : Focusing on factors that show the difference among non-frailty, pre-frailty, and frailty derived from this study can help in the intervention and management of preventing frailty by using a machine learning technique.
|
|
|
KEYWORD
|
|
|
Decision trees, Frailty, Lifestyle, Machine learning, Random forest
|
|
|
FullTexts / Linksout information
|
|
|
|
|
|
Listed journal information
|
|
|
|
|