Sunday, September 28, 2014

The Physician and the Autistic Savant

A person with autism who in mentally disabled and who has a low IQ but who is exceptionally gifted in a specialized field like mathematics.

Definition, autistic savant

The computer is an autistic savant, but it is not a person. It is an exceptionally gifted powerful took for processing information but outside of that information it is a moron.

This thought is not original with me. In 1967, Peter Drucker expressed it an essay, “The Manager and the Moron.” A computer cannot think for itself. It can only process information submitted by a human being, and the human being is the only one who can usefully use or interpret that information.

According to a new book by Walter Isaacson, The Innovators: How a Group of Hackers, Geniuses, and Geeks Created the Digital Innovation,” we are still far from the “thinking” machines envisioned by computer pioneer Alan Turing (1912-1954). Turing was considered the father of computer science and artificial intelligence who insisted computer would ultimately be able to “think.”

Isaacson should know the limits and power of innovation. He is chief executive of the Aspen Institute and has authored biographies of Benjamin Franklin, Steve Jobs, and Albert Einstein. Isaacson’s position is this: Today’s biggest innovations spring from a combination of human creativity and computer processing power.

Unfortunately, a mindset is emerging out there in the field of medicine and health care that computers can “think” for physicians and can dictate what needs to be done for patients based on computer-generated megadata and expressed in computer algorithms and protocols. This data, theoretically, even when blindly applied, will improve the overall “population health” of Americans.

This may be. But there is a much more likely possibility “that the combined talents of humans and computers, when working together in partnership and symbiosis, will indefinitely be more creative than any computer working alone.”

IBM, with its computer Watson, is pursuing this symbiosis by working together with oncologists as partners to develop more rational approaches to the treatment and cure of cancer.

I have long believed that the computer is useful in providing physicians with information they can consider, use, or reject based on their knowledge of the patient. Thirty years ago, with the help of Russell Hobbie, PhD, a professor of physics at the University of Minnesota, I developed a computer program that generated a differential diagnosis based on a patient’s gender, age, and comprehensive battery of laboratory tests. Our list of diagnoses included the probable diagnosis 80% of the time.

In today’s computer environment, this percentage would easily rise to over 90% given the information available on computer apps and electronic medical records. It would also be possible to calculate the patient’s level of health based in the patient’s physical measurements (weight, height, pulse, blood pressure, response to exercise, and key blood tests), all of which could be gathered during an office visit).

But I have one huge caveat about the use of this information. The physician, in concert with the patient, should be free to interpret or reject the computer-generated information. Artificial intelligence has limits. These limits should be respected or rejected in light of the patient’s emotional appetites, desires, drives, and instincts.

No comments: