Peter Szolovits, Ph.D.
Department of Electrical Engineering and Computer Science
Professor of Computer Science and Engineering
Professor of Harvard-MIT Health Sciences and Technology
MIT Director, Medical Informatics Training Program (HST)
Head, Clinical Decision-Making Group, MIT Laboratory for Computer Science (CSAIL)
Ph.D. Information Science 1974
California Institute of Technology
Our research centers on the application of artificial intelligence methods to problems of medical decision-making and the design of information systems for health care institutions and patients. We work on problems of diagnosis, therapy planning, computational aspects of genetic counseling, controlled sharing of health information, and privacy and confidentiality of medical record systems. We are particularly interested in applications in systems biology, such as investigating the relationship between genomics and clinical medicine. Several of our current projects address specific issues related to clinical care including home assistance to patients with chronic diseases, improved methods for monitoring intensive care unit data and genetic counseling.
Personally controlled health records
One challenge resulting from recent increases in employee mobility is the need to ensure continuity of care for individuals who change healthcare providers when they change jobs. At present, there is no single efficient method for exchanging health and medical records among current and/or former healthcare providers. Therefore, we are extending our previous work with the Personal Internetworked Notary and Guardian (PING), which is a distributed, web-based personally-controlled electronic medical record system. With our Boston-area colleagues, we are developing a New-England-wide data exchange network for medical data exchange that is based on an existing electronic data interchange network that was designed to track medical treatment and billing. We are now extending this network to include medical information in a standardized format with safeguards to minimize the risk to patient privacy. An essential feature of this system is access and control by the patients themselves, who are assuming more of the responsibility for coordinating their own care in complex cases. The goal of this work is to enable connectivity and interoperability within diverse medical care systems while ensuring both long-term access and confidentiality. We contributed to privacy provisions of the Health Insurance Portability and Accountability Act (HIPAA), and PING is designed to satisfy the requirements of this law.
Syndrome surveillance to detect disease outbreaks
With our collaborators, we are developing methods to screen hospital utilization data to detect patterns of symptoms or treatments that may signal a spontaneous or intentional outbreak of a serious disease caused by an infectious or toxic agent. This has two potential applications: 1) real-time epidemiological surveillance using computerized systems to collect data, communicate with other healthcare organizations and analyze patterns to identify emerging trends; and 2) developing models of normality, such as typical patterns of complaints and treatments that fluctuate in response to seasonal or episodic factors. The goal of this work is to detect new trends as they emerge within a given geographical area, even if we cannot identify specific patterns on a day-to-day basis. The current methodological challenges in this work include selecting appropriate methods to model patient volumes and complaints, defining a serious outbreak, and identifying people at risk in time to prevent or treat serious complications.
- Fraser HS, Szolovits P, et al. (2001). úTeleMedMail: free software to facilitate telemedicine in developing countries.? Medinfo; 10(Pt 1): 815-9.
- Riva A, Mandl KD, Szolovits P, Kohane IS, et al. (2001). úThe personal internetworked notary and guardian.? Int J Med Inf June; 62(1): 27-40.
- Mandl KD, Szolovits P, Kohane IS (2001). úPublic standards and patientsô control: how to keep electronic medical records accessible but private.? BMJ Feb 3, 322(7281): 283-7.
- Institute of Medicine (2000), Protecting Data Privacy in Health Services Research, National Academy Press.
- Kohane, I. S., H. Dong and P. Szolovits (1998). úHealth information identification and de-identification toolkit.? Proc AMIA Symp: 356-60.
Last Updated: April 16, 2008