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.

Incorporating repeated measurements into prediction models in the critical care setting: a framework, systematic review and meta-analysis

“…To extract features which describe the relation between already extracted features, various methods were described: multidimensional correlation analysis [11, 38], association-rule mining [29], sequence patterns of categorized variables [32, 36, 40], convolutional dictionary learning [52], the ratio between means in sequential periods [48], the number and duration of categorical variables under/above a pre-defined threshold [8, 33, 50] and cross-correlation patterns between multiple repeated measurement trends [7, 47].…”

“…by the student’s t test ( n = 10) [11, 28, 35, 38, 44]. Various other methods have also been described ( n = 6) [7, 29, 40, 47, 48, 52].…”

Rule-Based Models for Risk Estimation and Analysis of In-hospital Mortality in Emergency and Critical Care

“…AC methods have previously been used in various fields of healthcare, including the prediction of outcomes and adverse events ( 13 – 16 ), the prediction of diseases and wellness ( 17 – 20 ), as well as biochemistry and genetics ( 21 – 25 ). AC methods have also been used in the field of in-hospital mortality risk estimation using the results of 12 lab tests ( 26 ). This shows the feasibility of AC methods in in-hospital mortality risk estimation.…”

Table_2.XLSX

“…AC methods have previously been used in various fields of healthcare, including the prediction of outcomes and adverse events (13)(14)(15)(16), the prediction of diseases and wellness (17)(18)(19)(20), as well as biochemistry and genetics (21)(22)(23)(24)(25). AC methods have also been used in the field of in-hospital mortality risk estimation using the results of 12 lab tests (26). This shows the feasibility of AC methods in in-hospital mortality risk estimation.…”