Reengineering Patient Care with Artificial Intelligence

The decisions clinicians make often have life and death consequences. At the very least, they determine whether a patient receives a correct diagnosis or is sent down the wrong path, sometimes enduring years of needless suffering or wasteful medical testing. As the depth and breadth of medical knowledge continue to grow exponentially, the diagnostic and treatment options facing physicians and nurses have become so complex that it is sometimes impossible to make truly informed decisions. A recent editorial in the New England Journal of Medicine summed up the predicament succinctly: “The complexity of medicine now exceeds the capacity of the human mind.”¹ That perspective is supported by the statistics. By one estimate, a new medical journal article is published once every 26 seconds, which translates to about 5,000 articles per day. Similarly, the large amount of data in a single EKG tracing or CT scan can be overwhelming to analyze for a physician, especially in the short time frame in which they are expected to reach a decision. Even the data in an individual patient record may be more than many clinicians can manage, with pages and pages of information to wade through and very little time to arrive at the right diagnostic or therapeutic decision.

These problems become palpable when one looks at the statistics on diagnostic errors. A report from the National Academy of Medicine points out that about 5% of adult outpatients ...