ConfidentCare: A Clinical Decision Support System for Personalized Breast Cancer Screening

摘要

Breast cancer screening policies attempt to achieve timely diagnosis by theregular screening of apparently healthy women. Various clinical decisions areneeded to manage the screening process; those include: selecting the screeningtests for a woman to take, interpreting the test outcomes, and deciding whetheror not a woman should be referred to a diagnostic test. Such decisions arecurrently guided by clinical practice guidelines (CPGs), which represent aone-size-fits-all approach that are designed to work well on average for apopulation, without guaranteeing that it will work well uniformly over thatpopulation. Since the risks and benefits of screening are functions of eachpatients features, personalized screening policies that are tailored to thefeatures of individuals are needed in order to ensure that the right tests arerecommended to the right woman. In order to address this issue, we presentConfidentCare: a computer-aided clinical decision support system that learns apersonalized screening policy from the electronic health record (EHR) data.ConfidentCare operates by recognizing clusters of similar patients, andlearning the best screening policy to adopt for each cluster. A cluster ofpatients is a set of patients with similar features (e.g. age, breast density,family history, etc.), and the screening policy is a set of guidelines on whatactions to recommend for a woman given her features and screening test scores.ConfidentCare algorithm ensures that the policy adopted for every cluster ofpatients satisfies a predefined accuracy requirement with a high level ofconfidence. We show that our algorithm outperforms the current CPGs in terms ofcost-efficiency and false positive rates.