Artificial Intelligence in Population Health

Humans have been interested in artificial intelligence for as long as you or I can remember. If you’re a science fiction fan you’ll have seen it in both good (i.e. Star Trek’s Lt. Commander Data and Star Wars’ C3P0) and bad forms (i.e. The Terminator or Hal 9000 from 2001 – A Space Odyssey) since the 1980s. Suffice to say, we’re still a ways away from having Vision or Ultron, both from the Avengers, running around. However, AI has become the new buzzword in health technology because of its potential to reduce the risk of disease, help with decisions such as palliative care vs. hospice, and ultimately create overall risk stratification. AI requires data, and as the thirst for more data has grown in the past decade, the ability to filter out the “noise” has become a much more difficult task.

Tech hubs in San Francisco, Austin, and around the United States are frantically trying to figure out what algorithms will work to accurately predict patient behavior. This refers to what is known as deep learning, which is “a  type of machine-learning algorithm that is supposed to ape the neural networks of human brains, learning on their own to recognize patterns.” (Johnson, 2018). Essentially it’s Social Determinants of Health 2.0. We’ve successfully calculated human behavior in shopping habits. Ever notice how Amazon is always recommending other products that coincide with what you’re looking for? Have you noticed you tend to receive more coupons for the brand of dog or cat food that you usually buy for your pet? In both of these cases you have a retail center that is using AI to detect underlying patters in your shopping behavior and the company is trying to ensure you continue purchasing through them (and buy even more).

If we can create AI in the retail space, then this same type of AI can be used in healthcare. How often do you see your Primary Care Physician? Where do you see your PCP? When you have an acute issue, do you go to the Urgent Care or to the Emergency Room?  Do you have a specific day you tend to visit the doctor (maybe before work on a Friday)? All of these are patterns, and patterns can be learned.

Artificial Intelligence

It won’t be long before AI gets out in front of patients predicting who is most likely to end up in the ER because they aren’t seeing their doctor enough. There’s enough data out there to figure out someone’s eating habits and social habits. It’s a matter of harnessing it, and also allowing for the dots to be connected via interoperability. Your phone, the electronic health record at the hospital/doctor’s office, wearable devices, etc. All of these pieces will eventually come together. Last year Athenahealth merged with Virence Health Technologies to create a larger EHR platform. This year Tableau was purchased for $15.7 Billion by Salesforce to augment their data analytics platform. These types of trends in various segments of the healthcare industry will continue, and the result will be more data pooling. That’s a good thing for AI in the long run.

From a Value Based Care perspective this has the opportunity to make physicians’ lives easier (not harder like the first generation of EHRs did), make diagnosing more accurate, and ultimately drive down costs because of predictive analytics. In the beginning it was about getting the most data. Finally, we’re on the verge of using it correctly. It’s something that will take time, and that many companies say they can do however, it will indeed happen soon. CMS recently launched a competition for AI called the Artificial Intelligence Health Outcomes Challenge to determine hospital and skilled nursing facility utilization to prevent readmission rates. With financial incentives from the government and contract incentives from value-based payment arrangements, AI will surely come to the forefront faster than we ever thought. To use a quote from the Borg, another sci-fi artificial intelligence character, “Resistance is futile.”

Ryan Mackman

About the Author

Ryan Mackman, MBA, MHA - Business Consultant

Ryan Mackman has been an ACO business consultant team member with Salient since March 2018. In this role, he acts as a solution trainer, marketing and sales consultant, as well as Value Based Payment strategist. His skillset helps augment Salient’s efforts at the ACO and physician practice level.

Leave a Reply

Your email address will not be published. Required fields are marked *