The Future of Artificial Intelligence in Medicine
- by Jeff Voigt
Experts discuss how A.I. and evidence-based medicine will soon be an inevitable part of patient care
On April 10, I hosted a segment on Sirius XM channel 111’s program “The Business of Health Care” program. It focused on the use of artificial intelligence (A.I.) and its potential for accelerating evidence into everyday care. My guests were Dr. Craig Umscheid, Associate Professor of Medicine and Epidemiology at Penn and an expert in the use of evidence-based practices; Dale Van Demark, partner at McDermott, Will, and Emory, which specializes in the legal ramifications of A.I.; and Charles Chadwick, president of Live Circle, a home-based A.I.-driven care service for people with chronic conditions.
One of the greatest benefits of using A.I. in medicine is that it has the potential to unlock clinically relevant information hidden in massive amounts of data, which in turn can assist in clinical decision making. A second form of A.I., called machine learning, can learn from this data to improve its accuracy in decision making.
Evidence-based medicine is the second tool needed to make A.I. successful in health care. Simply put, evidence-based medicine is the synthesis of outcomes from high quality studies—ones that use meta-analytic statistical techniques. Those studies also form the backbone of clinical guidelines developed by specialty societies, and help determine what Medicare and private insurers cover and pay for. Artificial intelligence is being used to synthesize findings from studies in order to diagnose and help direct therapies for patients. A.I.’s use will likely continue to expand into numerous other areas in the future.
However, despite its positive trajectory, there are still key issues surrounding A.I. usage in medicine that need to be addressed: digital coverage that includes privacy and security; good data; reproducibility of A.I. decision-making published in high quality journals; coordination across disciplines and businesses; proper regulatory oversight; patient engagement; medical culture acceptance; health insurance modeling and payment; and allocating responsibility for the A.I. black box. While this is a long list of issues, the experts agreed that they will be addressed, albeit over time.
The last part of our conversation focused on an A.I. system being developed by Live Circle. Live Circle and care transition experts at Penn are exploring opportunities to build evidence around this new model. This learning system provides oversight for the care of chronic-care patients (who suffer from diabetes or early stage Alzheimer’s, for example) without overburdening their lifestyle and family. The focus of this system is ensuring that a patient continues treatment with minimal lifestyle disruption. The patient’s health and lifestyle is enabled in their home via coordination with caregivers, family members (even those remotely), nutrition, and transportation. The system is unique because it serves a patient’s life, not just his or her health. Some countries in Europe have already been implementing this system, and it is likely that long-term care in the United States will move this way in the future, as it continues to be driven by the cheapest options for patient care.