Artificial Intelligence May Lower Healthcare Costs, Propel Treatment Towards Precision Medicine

Edwards, a leading figure in the healthcare technology industry, discusses the impact of Artificial Intelligence and its abilities to cut healthcare costs, individualize treatment through more specific diagnostics and intuitive data, and much more in part 2 of an interview that he conducted with Pharmacy Times.

Q: Can you elaborate on Artificial Intelligence (AI) predictability and why it is important for individualized care?

John Edwards: Predictive models have to be trained on a set of data. And if the data is really narrow, it's really difficult to know whether the prediction that's coming from, and if the AI model makes sense. You know, only about 3% of the population participated in clinical trials. And so by its nature, clinical trials have a small group of people that drugs have been tested on.

And yet, when we put drugs out into the market, they get treated by everybody right there, the whole world starts to use the drug. But only 3% of that group had characteristics that led them to allow themselves to be part of the clinical trial. Real world evidence that can come from examining this data, and creating AI models driven from the data, allows us to use more data from a broader set of people—perhaps more representative of all of us. By offering more complete datasets across broader research [and] everyday evidence that can be collected, we can better understand what's driving the variations in the results and the outcomes from the medications that we give our patients or from, you know, the treatments that we choose to pursue.

Q: How are doctors and pharmacists granted authorization our data, and how is it used?

John Edwards: When we go in, almost every visit we will be asked to authorize (or will be asked to sign some things that say we're giving that provider the right to our information.) In Europe, you also have the right to withdraw your information, you know, through the privacy techniques— that doesn't exist so easily in the US. Once you signed that this treating doctor had rights to your information, they can use it within their institution. If the information gets shared with a payer health insurance company, they have right to the information so that they can administrate payments. So, in some weird ways, my insurance company has more rights to my collection of data than each of my doctors, because my doctors only have rights to the data that they collected while they treat it rather than my entire collection of data.

John Edwards: It should lower cost. Whatever we can do to get people to make appropriate use of providers, or pharmacy products, or diagnostic products that have been proven to work is the belief—it is the best path, or the cost is the most effective.

If we are off that path, and we're doing things that aren't effective, we're bringing detrimental results. We increase total costs because we're not offering people the best care for them in their conditions. And so, it could mean that some people's utilization goes up because they need more medications, or they need to see more doctors, whereas other people will feel comfortable following a protocol that may not require them to be seen as often.

Because their risk factors aren't there, that should cause them to be seeing so many doctors from a preventative medicine, or potentially taking medications that are not going to work for someone like them based upon biomarkers that have been discovered and then associated with previous people's outcomes (and you have the same biomarkers).

It takes us a step towards precision medicine. it starts to close the gap between the knowledge that's being produced about drugs and the practice of drugs with doctors, so they can better understand.

Q: How is AI individualizing treatment for conditions like breast cancer?

John Edwards: No radiation over a lifetime is measured. And if you have more and more radiation through x-rays, through various things that you have from a diet, you are taking on greater risks and the side effects of radiation. After you have cancer, you know, understanding the breast [and] where the tumor is, how dense your breast is, and where the placement is, causes decision to be made about how to apply radiation to the person's breast.

And yet those decisions aren't being run through an AI model that's learning from the body of evidence of all the breast cancer that we're treating it. It's done individually by that oncologist, and radiation therapist. So, we can do better. We can offer more complete data for decision sciences that then apply the capability. The tools are available to deal with. We just must be willing to build those types of solutions. and lean into the problem in a different way than we have [before].