5 Reasons Why the Computer Will Replace Most of What Health Care Professionals Do
The computer will replace most of what the health care team does.
Technology investor Vinod Khosla, who previously published the boldly titled "Do We Need Doctors or Algorithms?" article in TechCrunch, recently sparked controversy by challenging a room full of doctors to disagree with his argument that the computer will replace the doctor—a challenge that was met with silence.
I will go so far as to say the computer will replace most of what most of us on the health care team do, and I doubt I will face any more opposition than Khosla. Here are 5 reasons why:
1. The transition to automation has already occurred in other areas where we once thought human judgment was required.
Today, most commercial flying is performed via auto-pilot, and most stock market trading is performed by algorithms. Real estate agents, drivers, travel agents, and waiters also currently find themselves on the tipping point of a similar revolution.
2. There is increased focus on the "most pressing" symptom instead of holistic care, because there is less and less time.
A study of 29 family physician practices found that doctors let patients speak for only 23 seconds before redirecting them. In fact, only 1 in 4 patients were able to finish their statement. Although this study was published in 1999, no would claim that this has improved since then.
Looking ahead, it is easy to argue that it is only going to get worse. The "baby boomer" population continues to age, and the Affordable Care Act is predicted to add millions more insured Americans that need care, some of whom had seen doctors rarely, if at all, before.
The good news is that much of what healthcare providers do, including diagnoses, checkups, testing, prescriptions, dispensing, charting, and behavior modification, can be outsourced to (and performed better by) digital means of data collection, computers, and analytics.
3. We spend a lot, but we don't get much in return.
In a ranking of the most efficient countries for health care in 2014, the United States ranked 44th on cost and quality metrics. This should come as no surprise.
It is no surprise or coincidence that most medical decisions are made based on subjective determinants like habit, gut, ritual, instinct, experience, and anecdotal evidence. We cannot be expected to remember everything from school or to keep up with the nearly 2 million studies published every year.
The decision tree is perfect for a computer to model. How does Amazon know what I am going to buy before I buy it? How can I tell Siri “I feel like pizza” and she can tell me the best pizza places near me in my budget?
Should we not have something similar for diagnosis and treatment before we diagnose and treat? I would trust the health care baby of Amazon and Siri for an opinion far more than I trust that of a health care provider who's not looking at data to drive decisions.
4. We're about to be inundated with data.
It's not hard to imagine patients having more data than anyone else. Within a few years, a wristband will know a patient’s steps, sleep, heart rate, respiration rate, stress levels, metabolic rate, and other changes in body chemistry up to the minute.
We will also have a window into behaviors like medication adherence, and wherever this is stored, it will likely include a patient attitude/beliefs setting, offering choices gauging risk adversity towards treatments, feelings on end-of-life and intensive care, and even organ donor status.
5. 50% of us are below average, just compared with humans.
Computers are less prone to error and have the ability to get exponentially smarter over time.
In 2011, we learned that simvastatin 80 mg, a drug and dose that had been given to millions of patients for decades, has a potentially deadly side effect of myopathy, and we were asked to stop treating new patients with it. Today, most heart disease is identified only after patients have heart attacks. A computer enabled by machine learning that compiles enough data over enough time could identify trends and irregularities in data to optimize treatment and shift our focus from treatment to prevention altoghter.