By Maureen Stanton, CISE
Researchers from the College of Engineering and Boston Medical Center (BMC) will use a three-year, $900,000 grant from the National Science Foundation to develop and pilot a health informatics system to predict patients at risk of heart disease or diabetes, and enable early intervention and personalized treatment.
“Our research vision is to deliver personalized healthcare, from prediction to diagnostics to population health management,” said Professor Ioannis Paschalidis (ECE, BME, SE), director of the Center for Information & Systems Engineering (CISE). “In earlier work, we demonstrated how machine learning could predict hospitalizations due to these two chronic diseases about a year in advance with an accuracy rate of as much as 82 percent, a significant improvement over existing risk models — such as the Framingham study for cardiovascular disease. Now, with our synergistic team of scientists and physicians, we are developing more robust predictive methods and capabilities for offering personalized recommendations that can guide physicians and patients as they make health-related decisions. This is a new frontier in medical informatics research and we expect it will impact medical practice.”
Diagnosing these chronic diseases requires complex sets of clinical and pathological data, which often are not comprehensive, consistent, nor up to date for treating physicians. The result is that patients at higher risk often don’t get needed treatment, while those at lower risk do, leading to poor patient outcomes and unnecessary costs. The new, interdisciplinary research collaboration will be led by Paschalidis, SE Head Professor Christos Cassandras (ECE, SE, CISE) and BMC Associate Chief Medical Information Officer and BU School of Medicine Assistant Professor of Medicine Rebecca Mishuris. They will develop a new generation of predictive methods based on supervised machine learning techniques that are interpretable, have higher predictive power, and can handle more data.
The researchers will develop an approach based on novel mathematical methods and the requisite algorithms where electronic health records and real-time health data — including wearable, implantable, and home-based, networked diagnostic devices – can be used to develop prediction analytics that anticipate future events such as hospitalizations, re-admissions, and transitioning to an acute stage of a disease. These predictions trigger personalized interventions, ranging from increased monitoring and doctor visits to optimized treatment policies adapted to each patient.
The researchers will also focus on enabling personalized treatment based on learning and optimizing treatment protocols for chronic diseases. “Protocols of this type are typically empirical, using a one-size-fits-all approach,” explained Cassandras, head of the Systems Engineering Division. “They assess the stage of the disease and adapt medication based on the stage but not the patient. By incorporating specific personalized interventions with recommendations, clinicians can, therefore, intervene before a patient’s condition reaches a critical phase.”
In addition to methodological and algorithmic development, the project will pilot the newly developed algorithms by integrating them into the electronic health record system at BMC and with 14 affiliated Community Health Centers.
“Personalized, predictive healthcare is the future of medicine,” said Dr. Mishuris. “Our research is geared toward providing clinicians with powerful, interpretable data to achieve that goal. Physicians are drowning in data and administrative processes. Our research approach will help physicians manage the deluge of clinical and patient data to make decisions in a more systematic fashion.”
- How Machine Learning Is Helping Us Predict Heart Disease and Diabetes
- Using Big Data, Machine Learning to Reduce Chronic Disease Spending
- BU-Harvard Team Wins $1.2M NSF Grant to Improve Women’s Reproductive Health using AI and Machine Learning
- Researchers Win NSF Grant to Advance Study of the Cocktail Party Effect