TL-PC: An Interpretable Causal Relationship Networks on Older Adults Fall Influence Factors

Abstract

Identifying the internal relationships in the data is the basis of data analysis and prediction. Traditional statistics methods focus on testing the correlation of variables pairwise. However, the correlation has rather limited performance on real causal influence. In this paper, we focus on an interpretable and visible approach to detect causal relationship networks in order to study risk factors of older adult falls. Learning the skeleton of the network is challenging since it is hard to mine indirect relationships. Variables could have dependence given other variables. Furthermore, orienting appropriate direction is tough because real-world data may include hidden information. Researchers cannot control it like a simulated data set. Here we propose a method based on the Bayesian causal relationship, which we call the Time Logic PC algorithm (TL-PC). We use the TL-PC on the older adults fall application and show the explainable and reliable time logical causal relationships.