The objective of this project is to adapt condition monitoring technology for evidence-based care of psychiatric patients. Evidence-based patient care promotes the collection, interpretation, and integration of applicable patient-reported, clinician-observed, and research-derived evidence to improve the quality of clinical judgments and facilitate cost-effective healthcare. This philosophy is similar to that of condition-based maintenance (CBM) in manufacturing, which advocates the monitoring of production equipment using various sensors to enable real-time diagnosis of impending failures so the right maintenance actions can take place in a timely fashion and on an as-needed basis. The challenge for adapting CBM technology to patient care is mainly due to the large number of parameters and gaps (missing values) in patient data. A direct dimensionality reduction approach (as opposed to traditional dimensionality transformation techniques) will be used to identify and retain parameters that are necessary and sufficient for accurate diagnosis model building. An assumption free iterative imputation technique will be developed to fill data gaps. This technique will be benchmarked against extant imputation techniques and guidelines will be developed for selecting the best technique based on dataset characteristics. Diagnosis/prognosis models will be constructed using a knowledge-based modeling approach, which is an expert centered data driven approach that uses IF-THEN rules to integrate physician first-principles knowledge and knowledge embedded in data.
This work, if successful, will lead to the following benefits: (1) medication cost will be minimized because patient conditions are monitored to provide evidence for clinicians to optimize medication usage; (2) patient care quality will be improved because treatment and medication can be properly adjusted in a timely fashion based on changing patient conditions; and (3) patient safety will be maximized because medication side effects are monitored and can be kept in check. Lessons learned from psychiatric patient condition monitoring, diagnosis, and prognosis will shed lights on general implementation of evidence-based patient care. Specifically, the research will generate new knowledge regarding the integration of IF-THEN first-principles knowledge and data mining techniques for diagnosis/prognosis modeling in healthcare applications.
