Visual Analytics for Predicting Disease Outcomes Using Laboratory Test Results

Laboratory tests play an essential role in the early and accurate diagnosis of diseases. In this paper, we propose SUNRISE, a visual analytics system that allows user to interactively explore relationships between laboratory test results disease outcome. SUNRISE integrates frequent itemset mining (i.e., Eclat algorithm) with extreme gradient boosting (XGBoost) develop more specialized prediction models. It also includes interactive visualizations allow interact model track decision process. helps probe by generating input examples observing how responds. Furthermore, it improves user’s confidence generated predictions provides them means validate model’s response illustrating underlying working mechanism models through visualization representations. offers balanced distribution processing load seamless integration analytical methods representations support cognitive tasks. We demonstrate usefulness usage scenario exploring association acute kidney injury, using large provincial healthcare databases from Ontario, Canada.