Artificial Intelligence for EHRs, the paradigm shift and the cure for biased AI

The real question is, when will we draft an artificial intelligence bill of rights? What will that consist of? And who will get to decide that?
— Gray Scott

In my book, The Future of Healthcare: Humans and Machines Partnering for Better Outcomes, I spent a lot of time talking about how AI, machine learning and other technologies can revolutionize the field of healthcare. I also talked about it in my last article on Medium.

One of the most exciting new uses of AI — when it comes to healthcare professionals, at least — is the potential for it to be used to revolutionize our struggling EHR system. And it’s easy to see why.

I’ve talked before about how insane it is that doctors spend over two-thirds of their time filling out paperwork when they could have been spending that time with their patients. One recent report by Medscape, which is owned by WebMD, found that doctors spend an average of 13–16 minutes per patient. Meanwhile, many people are worried that their conditions aren’t serious enough to warrant professional help, so they keep their complaints to themselves instead of risking wasting their physician’s time. Patients are well aware of the shortage of time that doctors face — and those who were interviewed for the study talked about “the pressured context in which their consultations take place: the limited resources, the lack of time, and busy doctors.”

The balance is actually getting worse over time. Ten years ago, doctors ‘only’ spent an average of one third of their time on the paperwork. The problem is now so big that it’s having a serious impact on the level of treatment that patients can expect. As TheNextWeb reports, “In 2014, a health IT solutions designer named Jess Jacobs started keeping track of all the hours she spent at her hospital. She found that only 29% of her 56 outpatient doctor visits were useful. On average, she had to wait 20 hours to get a bed in the hospital. Other calculations showed that just 0.08% of her time being hospitalized was spent treating her conditions. Jacobs, who suffered from two rare diseases, passed away in 2016, which made her message all the more poignant.”


Milan Petkovic, head of the data science department at Philips Research, believes that increasing productivity and saving time is one of the biggest promises medical AI-solutions have to offer, explaining, “Doctors will provide better diagnoses and treatments with less time and effort, making healthcare much more efficient.”

The future of AI in healthcare is bright indeed, but there’s a long way to go before we get there. We need to solve the problems with today’s EHRs before we can even think about the sweeping improvements that AI could usher in.

Problematic EHR systems are exacerbated by prior authorization, which is effectively a rationing tool that insurance companies use to enhance their profits, and it does this by forcing patients and physicians to fill out a bunch of forms. These are submitted to the insurance company who — often after weeks of delays — will decide whether the insurer will reimburse the patient for the physician’s recommended treatment option.

When it comes to electronic health records, physicians and medical facilities are essentially forced to pay EHR companies for the privilege of inputting their patient data. That patient data can then be sold at a profit to the EHR company — but physicians and patients don’t see the benefit.