Study on Prediction of Activities of Daily Living of the Aged People Based on Longitudinal Data

The loss of Activities of Daily Living (ADL) is the most important health problem of the elderly, which is related to their quality of life and the nursing burden of family. Based on the survey data of China's elderly health impact factors (2002 ∼ 2011), the C5.0 method of decision tree was used to establish the prediction model of ADL. The variables of our model include demographic variables, personal habits, social activities, disease, mental state, and the current state of ADL. The experimental results show that: (1) The model can predict the ADL condition of the elderly people in the next 2∼3 years, and C5.0 has a higher accuracy than other methods; (2) The age is the most important factor, followed by the current ADL status; in addition whether to do house work frequently and residential area both have important influence on ADL status; (3) Doing housework, taking part in social activities and keeping good mental state are helpful to keep healthy.