Improving Personalized Clinical Risk Prediction Based on Causality-Based Association Rules

Abstract

Developing clinical risk prediction models is one of the main tasks of healthcare data mining. Advanced data collection techniques in current Big Data era have created an emerging and urgent need for scalable, computer-based data mining methods. These methods can turn data into useful, personalized decision support knowledge in a flexible, cost-effective, and productive way. In our previous study, we developed a tool, called icuARM- II, that can generate personalized clinical risk prediction evidence using a temporal rule mining framework. However, the generation of final risk prediction possibility with icuARM-II still relied on human interpretation, which was subjective and, most of time, biased. In this study, we propose a new mechanism to improve icuARM-II's rule selection by including the concept of causal analysis. The generated risk prediction is quantitatively assessed using calibration statistics. To evaluate the performance of the new rule selection mechanism, we conducted a case study to predict short-term intensive care unit mortality based on personalized lab testing abnormalities. Our results demonstrated a better-calibrated ICU risk prediction using the new causality-base rule selection solution by comparing with conventional confidence-only rule selection methods.

Figure 1

Flow of causality-based rule selection process. The process selects rules based on the ranks of confidence and causality values of all rules generated from personalized clinical conditions.

Figure 2

Flow of prediction assessment using newly proposed (i.e., confidence-causality-based) rule selection (CAU) strategy and conventional top-confidence-based rule selection (TC).

Figure 3

Calibration comparison results between the newly proposed confidence-causality-based rules selection (CC) and the conventional top-confidence rule selection (TC). (a) Percentage of CC and TC that are well calibrated. CC with different number of top-ranked rules all calibrate better than TC. (b) The percentage that CC calibrates better than TC with different number of top-ranked rules.