Identifying predictors of independence in toileting activities using machine learning

Introduction
Toileting independence is a key goal in stroke rehabilitation, yet no consensus exists regarding the factors influencing its achievement. This study identifies predictors of toileting independence in stroke patients using supervised machine learning with a random forest algorithm based on a multidimensional dataset.

Material and methods
The analysis used medical records from 30 stroke patients. The dataset included basic attributes (1 item), physical and cognitive functions (7 items), and laboratory tests (15 items). Toileting independence was classified into two categories, independent or dependent, as determined using machine learning.

Results
The random forest model achieved 75% accuracy in predicting toileting independence. Five factors were identified as significant predictors: the Hasegawa Dementia Scale-Revised (HDS-R), 6-minute walk test (6MWT), Berg Balance Scale (BBS), albumin levels, and age. These results indicate that cognitive function, lower limb performance, balance ability, nutritional status, and age play critical roles in achieving toileting independence.

Conclusions
This study highlights the multidimensional nature of toileting independence, emphasizing cognitive, physical, and nutritional factors. The findings can guide rehabilitation strategies tailored to individual needs. Furthermore, the application of machine learning demonstrates its potential to uncover complex patterns, offering a robust framework for improving rehabilitation outcomes in stroke patients