Prediction of Frailty Grade Using Machine Learning Models

Abstract

Nowadays, frailty is becoming a major issue for the aging population. Frailty grading is important for patient quality of life because it is a geriatric syndrome of decreased physiological reserve that leads to increased susceptibility to physical stress factors and susceptibility to cardiovascular diseases. In this context, this study seeks a solution to the fragility rating regression problem with K Nearest Neighbor, Decision Tree, Random Forest, Extra Trees and CatBoost models using time domain features extracted from the triaxial accelerometer signals collected during the TUG test. Moreover, estimating the grade of frailty involved in addition to diagnosing people with the disease will provide physicians with more detailed information about the patient and allow accurate and effective treatment/supportive treatment.