5 December 2019

AI can reduce hospital liability claims

Artificial intelligence can save lives in hospitals by helping doctors make better diagnoses, according to Harry Hoar, a doctor at Baystate Health.

Hoar was speaking at Cayman Captive Forum, on a panel entitled Can We Leverage AI To Reduce Professional Liability Claims? The answer to the question presented to the panel was a resounding yes.

Hoar said: “This is not about a machine doing the job for you. The human brain is better at some things and computers are better at some things. But computers can escape the biases that affect humans and they augment our ability to make good decisions.”

AI will not replace human physicians, he said. “Physicians who know how to use technology will replace physicians who don't.”

Being a good physician is no longer about having all the answers, he said, but following a process and collaborating with other doctors and technology. Machines are better than people are following processes, he added.

Nearly half of all diagnostic errors (46 percent) are a combination of system-related and cognitive failures, according to Hoar, while 28 percent are pure cognitive failures. Nearly one in five errors is system-related, with only 7 percent categorised as “no fault” errors.

The data show that too often doctors have the information available to make the right diagnosis, but through pressure of time, fatigue or other factors, make incorrect ones. In 2015 data show there were 18 million diagnostic errors per year. Diagnostic errors led to 10 percent of hospital inpatient deaths, or around 74,000 deaths. Around 12 percent of hospital adverse events are caused by diagnostic errors, Hoar said.

Hoar demonstrated how symptoms can mislead doctors into making incorrect diagnoses, and how AI can assess the same symptoms and suggest other diagnoses worth consideration. “If AI can prevent even a couple of major misdiagnoses per month, which I think is a conservative estimate, it will save hospitals from a few lawsuits,” he said.

One person in the audience suggested doctors could feel compelled to run endless tests to rule out possible diagnoses suggested by AI, pushing up costs and wasting time, but Hoar said Baystate has seen no evidence of this.

Lynn Tenerowicz, chief risk officer at Baystate Health Insurance, Baystate's captive, said physicians still make final diagnostic decisions, and using AI did not alter the hospital's liability, it only reduced the risk of incorrect diagnoses.