The decision-making capabilities of a primary care physician are generally fixed within a certain scope of knowledge; it’s simply impossible to maintain a specialist’s understanding of every disease state. Artificial intelligence (AI) has the potential to change that, enabling primary care physicians to advise patients on issues normally referred to specialists, according to Kamal Jethwani, MD, MPH, Senior Director, Partners Connected Health Innovation. But how can providers trust that the underlying calculations informing AI-aided diagnostics actually add up?
The answer, Jethwani told an audience at the recent American Telemedicine Association annual conference, may be “explainable AI,” which can clarify for clinicians what variables supported a particular diagnosis. He and his team are investigating how enhanced transparency into AI technologies’ underlying algorithms can enhance provider trust in such technologies’ capacity for sound decision-making—and ultimately increase their adoption.
During a panel discussion at the conference, Jethwani also proposed that medical schools could integrate AI units into their curriculum to normalize the use of these clinical aids for future practitioners. The payoff, he noted, could be immense: In just one example, his team is investigating a system combining Fitbit data with electronic health records to predict with 80 percent accuracy when patients with COPD will experience an exacerbation of their condition.
For more, click here.