The Transformative Potential of AI to Enhance Patient Treatment

AI is being utilized to prioritize clinical trials by identifying areas of significant unmet medical need, such as Alzheimer’s disease, ensuring resources are directed where they are most required rather than toward well-treated conditions. A digital representation of a patient comprising genomic data and therapy history may be used to facilitate precision prescription, helping clinicians select the optimal therapy from a pipeline of candidates based on an individual's unique biological differences, according to Michelle Longmire, CEO and Co-Founder, Medable. Beyond the lab, AI can streamline the "final mile" of medicine by educating physicians on newly approved drugs directly within patient records, bypassing the need for doctors to manually track every new approval.

The introduction of agentic AI aims to solve the inefficiency in clinical trials, where currently 80–90% of processes are manual. By deploying autonomous agents to handle tactical data collection and "heavy lifting," AI removes the administrative overhead that currently occupies half of a professional's day. This shift allows human experts to focus on strategic work, such as optimizing site openings or trial enrollment. Longmire predicts that by 2030, while technology will consume over 60% of budgets, the overall cost of healthcare will decrease.

Pharmaceutical Technology® spoke with Longmire during the JP Morgan Healthcare Conference, being held in San Francisco from January 12–15, to find out more about how AI and machine learning are improving the use of patient data to enhance treatment options.

PharmTech: How can AI/machine learning be used to predict which therapies will best meet treatment needs, especially for rare diseases or cancer treatments?

Longmire: [Approaching] the problem from post-discovery and looking at the in-human trials, the most important thing that we really look at is how can you accelerate the inhuman testing part? When we think about unmet needs, we're looking at it from [for example] you've got 9000 candidates or a number of different treatments you want to try and [put through] clinical trials, where is the real unmet need? So that would drive the prioritization in the resourcing and timeline of the clinical trials. [If] you have 10 or 20 different candidates, and one is for psoriasis, and one's for Alzheimer's disease, I think, from our perspective, there's already great therapies for all psoriasis. So, let’s focus on where there really is an unmet medical need, and target company resources towards that need first.

How are AI and ML platforms being utilized to match the patient to the treatment?

We know that certain people respond to certain therapies better than others. One of the things that we've really been advocating for is something we call the digitome, which is a digital representation of a patient, and this enables us to get more precise, better data. This would include genomics data past therapies and failed therapies, in some cases, even looking at the genomics of drug processing. One of the areas that we're really excited about for machine learning and AI is, how do you do precision prescription? Let's say AI discovery unlocks the pipeline, and you've got 10 different therapies for a disease. What's the best one for the patient in front of me? That's a really important role for AI. And I think it's going to be exciting as more people have high-fidelity data about what makes them different as humans versus what makes us all similar. Clinical trials are good at solving for the common denominator in humans which is key because, of course, humans are more alike than different. But when prescribers can turn to many options, it is better to take a precision prescribing approach to decide which of these different therapeutic options is best for an individual person based on their differences.

What other uses for AI or ML do you see on the horizon?

Generally in healthcare, there's a big opportunity to use AI to do some of the initial triage, even diagnosing and prescribing as described earlier. I am a trained dermatologist in a relatively small pool of about 8000 dermatologists in the United States, which means there is often a two- to three-month wait time for patients to see one of us. Way back in 2013, Medable had a patent on using machine learning and computer vision for visual diagnosis. We have the data and publications galore that show this is safe and effective. It's just not widely being used. As people become more comfortable with this and we have reimbursement models, then we will be really able to target these unmet needs. AI could even be leveraged for initial triage, diagnosis, and prescribing in specialty care areas.

How do you think these platforms can prevent the bottleneck that happens with drug discovery, drug formulation, or just getting a drug to market?

One key way to accelerate commercialization is by educating physicians about new therapies. Medable and the industry invest a lot of time and money in driving new therapies to approval, but honestly, most doctors aren't even aware of cutting edge therapies available. There's a big opportunity to use AI to help identify the latest and greatest drug for each individual patient. And take some of that final mile work [of] “hey, have to learn the last 54 drugs that were approved last year, or the five that relate to my specialty,” off of a doctor’s plate and streamline that for physicians. It empowers them to help their patients sooner rather than waiting to hear about some new breakthrough at the next the big medical conference in the year.

What are you presenting at the JP Morgan Healthcare Conference this year?

We're really focused on the fact that 80–90% of clinical trials are manual, and that, in our opinion, this is a big reason why we haven't changed the number of drugs approved in a year, no matter how much technology we throw at this, because you're only really previously looking at 10–20% of where you could apply technology. Now with agentic AI, we can remove that tactical overhead. We can actually do many more complex things with AI that were formerly only able to be done by humans. We've released agents that essentially autonomously do significant parts of the work to enable humans to do more strategic work and remove that manual bottleneck.

How do we end up in a world where there's real autonomy to AI to do a lot of the heavy lifting, and humans can do the more strategic work? Most of the [people] we've spoken with as we've developed the technology, about half of their day, if not more, is spent doing tactical data collection, you know, going to the system, getting this data here, putting it in an Excel sheet. A minority of time I spent looking at that and saying, okay, strategically, what do we do to make this trial enroll faster, to open this site faster, to address the fact that say there aren't enough patients meeting the inclusion exclusion. So, we're using AI to remove that whole tactical overhead. And we envision that, you know, by 2030, probably 60% plus of the budgets will go to technology, but the important part, the overall cost will go down, and humans will play a much more strategic role.

What are regulators like FDA saying about this use of AI?

I think their guidance is at a fairly high level. But the really positive thing is, most of these have a good clinical practice lens, which is just a set of guiding principles. It's not prescriptive. So, there's a fairly strong framework for good clinical practice and being able to show that AI does good clinical practice as well as humans or better.

About the speaker

As the co-founder and chief executive officer of Medable, Dr. Michelle Longmire is mission-driven to accelerate the development of new therapies for disease. A Stanford-trained physician-scientist, Dr. Longmire witnessed firsthand the critical barriers to drug development – including the time and costs associated with clinical trial participation. She founded Medable to pioneer a new category of clinical trial technologies that remove traditional roadblocks to participation and radically accelerate the research process. Medable is now the industry leader in decentralized and direct-to-patient research, with the ability to serve patients in over 120 languages, 60 countries, and across all therapeutic areas. In addition to having raised over $500M in venture capital and driving Medable to an industry-leading position, Dr. Longmire has received recognition as a leading innovator and businesswoman, including being named as one of the 100 most creative people in business by Fast Company.