Is it Time to Replace Doctors with Artificial Intelligence?

May 24, 2017

Hey Siri, how far is the sun from the earth? Alexa, what does the weather look like today? When I ask Alexa to play some music, “she” responds with “here’s a playlist you might like.” When I buy something on Amazon, the site quickly shows me items that are “frequently bought together”, or suggests products that other customers like me bought in the past.

But how about this for a scenario?

Google, how does my health look today? Hey Siri, I’m having blackouts and I feel like vomiting. Siri or Alexa would respond with questions like, “Do you sometimes vomit shortly after a meal? Do you have diarrhea? Do you have heartburn? Do you have indigestion? Then based on my answers, I’d be told what the problem or medical condition is, and that I should take this medicine or go for that lab test. Wouldn’t that be amazing?

No longer would you have to visit a doctor for the same advice you could get sitting at home. And there would be no more worries about insurance or waiting for an appointment. You could simply discuss your symptoms, and a personal assistant or voice bot like Siri, Google, or some other mobile or web-based app would recommend customized treatment, and continuously monitor your health. The app would also provide all the information you need to stay on top of your condition, based on your habits, symptoms, and other relevant factors. Just like a personal doctor.

Sound good? That’s the scenario I’m going to discuss in this article.

How exactly does technology communicate with us when it takes our shopping orders, answers our questions, and recommends what we might like or dislike? The answer lies in Artificial Intelligence (AI).

But what is Artificial Intelligence? Let me give you some examples instead of going into a technical definition.

H&R block developed a software program that asks tax-related questions about events that could lead to a tax deduction — things like selling a house, or buying shares. You no longer need a human accountant for tax filing. You can get tax advice by sitting at home and interacting with software.

Google Maps can track your destination history, including the time of day and specific locations you typically drive to and from. Maybe you take the same route to work every day. Google Maps will automatically guess you’re going to work at a certain time, and will begin to tell you the quickest way there — particularly if there’s traffic.

You can switch your lights and TV on and off, and open and close your garage door, simply by using a phone app, or issuing a voice command to Alexa or the Google Home assistant. You can even request the latest lame joke, if that’s what you’re into.

Siri, Alexa, and Google Home controller are all examples of chatbots and voice bots that are driven by artificial intelligence. Tools like these have become an integral part of our lives, and Gartner predicts that by 2020, “the average person will have more conversations with bots than with their spouse.” Apparently when we find a way to ignore our spouses, we call it “innovation” 🙂 .

But kidding aside, AI tools not only give us the ability to consolidate our daily habits and activities — think Apple watch or Fitbit — they can also provide ongoing guidance in examining our health, and predicting if we might be at risk for diabetes, heart attack, cancer, or some other chronic disease. They can even suggest early prevention steps that might help us avoid such risks.

So, how does it all work? Every request that you make through a website or app gets stored in the form of data. Algorithms then analyze the data and identify a pattern, so that if the same request is made again, the correct answer will be provided. These algorithms can also learn from their mistakes.

Leveraging AI-driven web applications and electronics in the medical domain

I was listening to the Elvis Duran show the other day, and Elvis was suffering from travel constipation. He asked Siri to call Dr. Oz and find out which medications might help him to find some release :-). Siri called, but because Dr. Oz didn’t answer, Elvis didn’t get the information he needed. What if Siri could have recommended that Elvis take “XYZ” medicine, or had been able to provide some of the other advice that Dr. Oz might have given?

Let’s look at some of the exciting ways that different types of AI-powered tools have been injected into the medical domain.

Medical Chatbots — Chatbots are nothing more than the Siri that chats with you and responds to your questions. Chinese search engine giant Baidu is launching a medical chatbot designed to make diagnosing illnesses easier. The conversational bot is named Melody, and it comes built into the company’s iOS and Android Baidu Doctor app, which launched in China in 2015. Baidu Doctor allows users to contact local doctors, book appointments, and ask questions, with the chatbot helping to speed up the process.

Automated Online “Symptom Checkers” — These tools give patients a quick opportunity for self-diagnosis. Tools like WebMD, Mayo Clinic’s Symptom Checker, and Health Net generate a medical questionnaire from the demographic information and symptoms you provide. Based on your response to this questionnaire, the apps then provide a list of the most probable diseases you’re likely to have. Some symptom checkers even suggest recommended treatments.

So, could AI tools do away with physicians?

While initiatives like these would seem to provide great evidence that yes, machines can replace your doctor, the answer to this question is still a resounding “NO”! And that’s because, as Spiderman once said, “with great power, comes great responsibility.” Artificially intelligent machines may have power, but they lack responsibility. When such tools provide a recommendation, they don’t see the patient or feel their problem like a doctor does.

A doctor’s expertise is invaluable because:

  1. Medical professionals have experience that comes from treating multiple patients over many years
  2. They understand the complexities of various scenarios that would take years of time for a machine to learn
  3. In an emergency, you don’t want to depend on a machine for help when you need to reach a hospital ASAP
  4. Patients are much more comfortable taking treatment recommendations from a doctor than they are from a machine

Technology Gaps

In one recent study, online ‘Symptom Checkers’ were found to misdiagnose more often than not. Researchers tested 23 online symptom checkers, and found that the correct diagnosis was provided first on a list of potential illnesses only about a third of the time. That means that symptom checkers are spitting out wrong diagnoses the other two-thirds of the time. “People who use these tools should be aware of their inaccuracy and not see them as gospel,” says Dr. Ateev Mehrotra, who led the research and is a professor of health care policy at Harvard Medical School. “They shouldn’t think that whatever the symptom checker says is what they have.”

Symptom checkers are based on medical dictionaries, and don’t think for themselves or use professional judgement like real doctors do. You simply can’t take what they say as the final word, and rush off to the emergency department. These tools do get updated, but there are many data points that must be considered before an accurate diagnosis can be provided to a patient — including things like lab test results, daily habits, and recent health episodes. It’s quite impossible for a machine to consider your health concerns from the various angles that a doctor does.

So, where does that leave us?

I wouldn’t recommend that you become completely dependent on AI-driven technology tools for making medical decisions, since this area still requires a lot of research and development. But I would suggest a focus on building out AI tools that focus primarily on helping physicians gather insights for making better decisions, and customizing treatments. One of the areas in need of such a tool is Primary Care. Access to tools that could be used during patient checkups would save a great deal of time, since physicians could use them to collect patient profiles, and find relevant recommendations based on historical data. The doctor would then be in a position to validate this information and provide personalized treatment.

Doctors have become so busy that it’s very difficult for them to provide customized treatment to each and every patient over the course of a regular checkup. I remember an examination where my doctor checked my blood pressure, weight, eyes, stomach, and then asked me if I had any problems. I said yes, I’d had a headache five or six times over the past month, and the physician said it might be viral. There are many chronic diseases that start with a headache. If I’d had one of those, I wouldn’t have known it until I started to see some serious signs. The bottom line is that many studies suggest medical errors are now the third leading cause of death in the U.S.!

In many cases, chronic diseases like diabetes and cancer do not develop overnight. They show symptoms, but the average person without medical expertise often ignores these signs, or discounts them as unimportant since they’re often just run-of-the-mill indicators like headache, vomiting, or fatigue. We treat our symptoms with home remedies and over-the-counter medications, even though they may be the early warning signs of serious future risk. The real question here is, how likely is it that I’m going to remember them all by the time I have my next regular check-up?

That’s where artificially intelligent tools can be helpful in measuring the potential risk of any chronic disease that starts with normal symptoms. They allow us, or our physicians, to input all our symptoms at the time that they happen, then provide a recommendation, or show risk probability, based on other patients with chronic diseases that started with similar complaints. This can help us to avoid serious future health risks outright, or at least get started on early treatment.

My conclusion is that artificial intelligence is NOT going to replace doctors. But it can be leveraged in specific areas like the normal patient check-ups that are part of primary care. As a starting point for giving physicians insight into what may be wrong with their patients, AI-driven tools could prove invaluable for enabling doctors to provide the help that’s really needed. And that can lead to more timely decisions, and fewer medical errors.

This article has been re-published from author’s LinkedIn post. For reference – Click Here.

Shail Choksi

Shail Choksi

Senior Consultant

Shail is currently a senior consultant for Orion System Integrators, LLC. He support all phases of strategy consulting engagements focused on Data Analytics and Business Analysis. In his early career, he started an e-commerce business, making him the youngest IT consultant for the state government of India. Among the awards and accolades, he was the recipient of the Rajiv Gandhi award for Youngest Achiever in IT from the All India National Unity Conference. He has also been nominated for the Godfrey Philips Bravery Award, for outstanding achievement in innovation. His research on professional accreditations was published in the American Journal of Business Education as well.

Shail holds a Bachelor in Commerce with Accountancy from Gujarat University, India and an MBA in Finance from New York Institute of Technology.

Subscribe to our Blog

Get updates on new articles, webinars, whitepapers and other resources.