Pitt Clinical AI Drives Patient Safety Research in JHF Initiative

Medical errors kill more than 250,000 people annually in the United States, according to a 2016 report by the Johns Hopkins School of Medicine. More recently, spikes in medical errors were reported in 2022.  

Since the Institute of Medicine’s 1999 landmark report “To Err is Human” estimated that tens of thousands of deaths occurred annually due to medical errors, interventions at health care institutions across the country have tackled medical errors. Estimates showed that many interventions produced short-term improvements, but the improvements didn’t last. Medical errors are getting worse, not better, and COVID-19 compounded the situation by creating overworked and burnt-out providers. 

Pittsburgh’s Jewish Healthcare Foundation (JHF) has worked for more than 25 years developing patient safety interventions that have become regional and national models. The JHF has now created an ambitious plan to bring together the medical, AI and robotics expertise among Pittsburgh institutions to develop autonomous patient safety technologies. Called the Regional Autonomous Patient Safety (RAPS) Initiative, the project aims not only to reduce medical errors but to put Pittsburgh on the map as a global hub in a new industry.

The University of Pittsburgh is at the center of one of the first efforts funded by RAPS. The Medication Error Avoidance at Region Scale (MEARS) study is a collaboration between the School of Medicine’s Department of Biomedical Informatics, School of Pharmacy and School of Public Health. MEARS aims to reduce preventable adverse drug events that occur to patients who transition between skilled nursing and other care facilities. It will also develop and pilot a decision support system that brings together multiple forms of data – such as medications recorded in a patient’s electronic health records – to train an autonomous AI-based technology that comes in an easy-to-use format for providers.

Richard D. Boyce, associate professor of biomedical informatics and principal investigator on the MEARS project, thinks the project is particularly important in view of a national aging population that may be cared for in areas like skilled nursing facilities.

“Oftentimes a patient in a long-term care facility goes to another facility for treatment or therapy, and medication errors occur during those transitions,” Boyce said. “For example, somebody has a medication that has helped manage their depression and the medication is accidentally overlooked in the transition and no one detects the problem. That can badly set the person back.”

“Transitions present a clear point to intervene to create automated detection of these medication errors,” he added. “But the technology is only one part – we need to coordinate all the human systems to answer questions, like who needs to receive the alert at what place in the workflow, and what is the best mechanism for them to receive the alert.”

Michael J. Becich, associate vice chancellor for informatics in health sciences and chair of the Department of Biomedical Informatics, believes that the RAPS initiative makes Pitt central to a regional collaboration which aims to scale up nationally in partnership with the Jewish Healthcare Foundation.

“This is a technology that can greatly impact the region of Pittsburgh,” Becich said. “It is a recognition that a lot of the artificial intelligence (AI) and machine learning talent in the country is concentrated here. We are marrying the Pitt Health Science biomedical expertise with AI, machine learning and robotics at Pitt and CMU.”

The AI and machine learning systems require data on which to train, and Pitt’s Department of Biomedical Informatics can call upon a vast amount of health data.

Becich believes the impact of the RAPS projects can grow far beyond research. Investment funds are coming from the federal government as grants regional demonstration projects and those projects are attracting attention from venture capital firms in the private sector.

Kay Metis, executive director of programs and strategic initiatives in the Department of Biomedical Informatics, considers this kind of data expertise an untapped resource.

“AI technology is underutilized within patient safety innovations,” Metis said. “There have been smaller groups that have run research and testing with promising results, but we want to see more investment. We’ve got all the right pieces here in Pittsburgh to really do some amazing work. Research and innovations coming out of here could have not just regional impact but really have national and global implications.”