Profile
- Herbert Chase, MD
-
Professor of Clinical Medicine (in Biomedical Informatics)
Director of Health Information Technology Certificate Training Program
Clinical decision support, pharmacovigilance, information retrieval, education
DBMI - 622 West 168th St., VC-5
New York, NY 10032
Phone: 212-305-1087
Email: herbert.chase@dbmi.columbia.edu
Website
Publications (Pubmed) - Education
-
BA, 1970 Brown University, 1966-1970
MD, 1974 Mount Sinai School of Medicine, 1970-1974
MA, 2009 Columbia University, 2006-2009
- Research Interests
- CKD-CDSS Project: There is an epidemic of chronic kidney disease (CKD) in the US and worldwide. The key to prevention and retarding progression is early recognition so that recommended management can be implemented. We are developing two CDSS tools which will promote early recognition and documentation which will improve management and outcomes: a CKD-Notification Tool which identifies and notifies providers who have not documented their patients’ CKD in outpatient notes; and a CKD Quality of Care Tool which presents feedback to providers (how the patient is “doing� in relation to the recommended guidelines).
Summarization Project: Patients’ records, especially those with chronic illness, are exceedingly complex. There may be literally thousands of pieces of information contained within clinic notes, discharge summaries, laboratory data, and various reports. The Summarization Project seeks to create a patient summary, represented as a dashboard, from which physicians can get an immediate overview and status of the patients’ pertinent and active medical problems, current medicines, and other essential pieces of information. From the dashboard the provider can drill down to specific data and reports. The summarizer will extract both structured data, such as laboratory values, and unstructured data, such as disease status, that is mentioned in the narrative portion of the notes.
Pharmacovigilance Project: The literature is rife with reports of new drugs, with marginal efficacy, that have serious and sometimes lethal side effects not known at the time of FDA approval and release into the marketplace. It takes at least five years for most of the serious adverse events of new drugs to be identified. The pharmacovigilance tool will use NLP to “read� the notes of patients, query-based extraction of structured data, and statistical methods to look for associations between drugs and medical findings (such as elevated liver enzymes or “edema�). Strongly associated drug-findings pairs may represent adverse drug reactions, perhaps previously undocumented. -
Courses
Biomedical Informatics Course for Fourth Year Medical Students
