Computers Are Eating Healthcare, Part 2

This post was originally featured on HIStalk.

This is the second in a series of posts that outlines how computers are eating the world. Check out Part 1.

Humans do three things and only three things: process information, store information, and share information. Everything in life, both virtual and physical (firing neurons drive physical movement), can be understand in these terms. Humans never can and never will do anything else.

Computers do three things and only three things: process information, store information, and share information. Computers never can and never will do anything else.

Unlike humans, computers excel at jobs that are intrinsically repetitive across all three dimensions of performance: processing, storing, and sharing information. Computers process, store, and share at infinite scale.

Because computers excel at these repetitive functions, they’ve dismantled most repetitive administrative and information distribution jobs that aren’t propped up by politics or regulations. Computers were the empowering tool that led to the dismantling of Blockbuster, the travel agency industry, the newspaper industry, and many others.

Healthcare is intrinsically repetitive. Will computers eat healthcare too? If so, how?

In the first post of this series, I argued that computers will eat radiology first. Radiologists look for patterns in images and cross-reference the EMR. The job requires almost no patient interaction. Occasionally radiologists query the PCP. Computers can perform those functions better, cheaper, and faster than radiologists ever could. Radiologists are slowly acknowledging that fact. This explains one of the themes of the most recent RSNA conference: interact with patients. Computers are still years if not decades away from naturally conversing with patients and providers. Radiologists want to interact with patients after decades of avoiding patients to avoid being eaten by computers.

Sedasys has built a new anesthesia monitor with a unique twist. It intelligently and automatically administers anesthesia during surgical procedures. Anesthesia, like all other forms of medicine, is intrinsically repetitive. Intra-operative anesthesia is at its core a negative feedback loop. Computers can automate that process.

IBM’s Watson is attempting to do the same thing with diagnostics. We are probably a decade away from that, though. Automated diagnostics are extremely difficult for a number of obvious reasons: conversing with the patient and understanding context, asking the right questions, natural language processing, processing enormous amounts of data, and many others.

Airplanes have been flying themselves for over two decades. With airplane black boxes, airlines have second-by-second records of millions flights spanning every conceivable weather condition in every conceivable geography. Planes automatically adjust flight paths based on what they’ve learned from every flight since the inception of black boxes.

And yet we still have pilots. Why? Regulation, and to respond in case the computer fails. Although I can’t find data to back this up, it seems reasonable that pilots boost safety metrics from 6 sigma to 7 sigma. Computers are responsible for the first 6 sigmas.

Computers will eat providers just as they’ve eaten pilots. Providers should look to the airline industry to understand what that means. The airline industry is medicine’s harbinger in a computer-eaten world.

In the future, providers will monitor computers, just as pilots monitor planes. This begs the question: will providers tell a computer what to do, or will providers be told by a computer what to do?