2018: The Year Of Machine Learning And AI In Healthcare

By Daniel Kivatinos, COO and Co-Founder, drchrono

Companies, venture capitalists and accelerators like Y Combinator are investing in Artificial Intelligence and Machine Learning. There are large investments into deep learning from many different industries, such as the development of self-driving cars.

If you think about AI and how it will help the healthcare industry, it makes a lot of sense. There are only so many physicians per person in the world and having some assistance can bring down the cost of healthcare and help patients at their time of need.

Let's look at the patient journey. A person is at home and wants to talk with a physician about something. At home the patient can book an appointment via a phone call or app. Today patients go to the doctor, sit in a waiting room and fill out forms. Then they see the doctor and leave the office, and there is follow up that might happen from the visit.

Below is a small list of some of the ways that AI can help with the patient journey.

1. At Home: Do you know the difference between all of the medical provider types there are? According to AACM there are 120 different medical specialties. There are a slew of other types of providers outside of this list as well. AI can help determine what type of provider or specialist you should go to based on a series of questions you answer.

2. In the Waiting Room: In the waiting room, via a check-in app, AI can ask 20 questions, and guess what you are thinking about. If you have 20 yes or no questions, AI can create more questions to ask more focused questions based off the initial 20 questions. For example, with the question, "Do you cough?” the amount of question options grow exponentially. If you ask one question, you can rule out a bunch of others. Having an AI intake app helps steer the questions and get focused answers from a patient, not just using a static questionnaire like the way things are today in the waiting room, when medical staff hand you a paper clipboard to fill out this information.

3. With the Doctor at the Point of Care, Diagnosis: If a patient is referred from another doctor, AI can standardize the way the doctor reads notes from different formats, presenting the information differently depending on the doctor and how they prefer to see information laid out.

Some doctors will never stop writing on paper or they might be using iPad with Apple Pencil. When natural language processing is good enough, the doctor will be able to put the paper medical record in a scanner and AI can read off what is on the paper and put it into a structured format that can be placed into a digital medical record. AI might also be able to extract text from a PDF, take that natural language and have it organized.

AI will start to help diagnose patients, learn from large data sets, and help us understand more about healthcare at a deeper level. Software can get better results at pattern recognition than humans by using deep learning. For example, in the near term future, AI might be able to read CT Scans and X-Rays better than a radiologist. Facebook can determine whose face is whom, and search engines can now distinguish pictures of the body sections, such as the difference between pictures of feet and faces. Sometimes doctors don’t recall a patient, but with an iPhone X there is now facial recognition. If facial recognition technology is leveraged in the future, just one snapshot can give you insights about a patient.

Also, after years, doctors build out really big patient files. The doctor might remember a patient but after 65 years of health records data collected they might not look through all of the data and miss something important. But AI can look and see symptoms and relate it back to something that happened a long time ago, perhaps something that was dormant and would be hard to find in a file.

I see AI assisting providers, giving insights into what is happening with the patient.

4. Billing the Patient: There are also other areas where the U.S. and the world spends massive amounts of money, like in medical billing. Generally companies have vast teams with large amounts of knowledge who spend countless hours trying to get payment for a medical practice.

With AI, claims adjudication might just become a bit more streamlined. Claims adjudication is a phrase used in the insurance industry to refer to the process of paying claims submitted or denying them after comparing claims to the benefit or coverage requirements.

5. Patient Follow Up: Some patients listen to the doctor’s advice and some don’t. AI can help with patient adherence to medications, log moods and do various other tasks. For example if a patient is to take a medication say 4 times a day and record how they are feeling, sometimes people forget to follow up with “doctors’ orders.” An AI bot can become part of the provider’s care team and do a daily check in, ask a few questions to keep the patient engaged, and help push them to adhere to the providers’ orders, perhaps by sending a friendly email with a human like question or a text or phone call.

Summary

I like to think about AI as a copilot, helping assist patients and doctors along the patient care journey with the provider and patient in the driver seat. Also as computing power is advancing combined with specific bleeding edge chip design, optimized algorithms, and mountains of data available in the cloud, there truly is a new momentum within deep learning and healthcare.

About the Author:

Daniel Kivatinos drives the direction, brand vision, and strategy for drchrono. Daniel's focus has been in the technology space since 2001, as a software engineer, designer and entrepreneur. Daniel’s goal is making healthcare simple and bringing innovation to healthcare with drchrono, a platform for physicians and patients