How Power, DPHARM Idol 2024, Is Building the AI Platform for Central Patient Recruitment
Power won DPHARM Idol Disrupt 2024 with its AI-enabled simplified platform that connects patients and sites to clinical trials. Brandon Li, Power’s Co-Founder and Co-CEO, tells us where the company is one year later and how it is disrupting medicine development.
To read more interviews with DPHARM Idol Disrupt winners, click here.
How is Power changing the way we do clinical research?
Our work is going a layer deeper on patient recruitment: finding a new way to engage patients, understanding who they are and what studies they’re qualified for, and making it easier for research sites to connect to the most qualified patients.
Why do you think Power’s platform won DPHARM Idol Disrupt?
It’s quite hard to marry a patient-centricity approach with an actual clinical operations challenge because a lot of patient-centric efforts are hard to tie to clinical operations goals. I think what struck a chord with the DPHARM Idol judges was seeing how we could build a platform that was truly centered on the patient experience of discovering research and turn that into a system that makes recruitment more effective for everybody involved.
How are you engaging patients in clinical research?
We built a patient-friendly website as a resource for learning about trials. We have our own AI models that scoop up the publicly available information and summarize it in plain English, so that patients can see the gist of a study, so that anybody could go on the website, learn what’s actually out there and who the leading medical researchers are that can help them. The net result is that we have thousands of people signing up every day on our website, who are interested in clinical research.
We always ask, “Why are you interested in research?” when anybody signs up on the website, because we’ve all seen the statistics about what percentage of patients don’t participate in research and how difficult it is for new patients to get involved. The responses we hear are so moving because they are from real people working through real medical situations, who are looking for help and understand that research might be an option for them.
"What struck a chord with the DPHARM Idol judges was seeing how we could build a platform that was truly centered on the patient experience of discovering research and turn that into a system that makes recruitment more effective for everybody involved."
How are you making sure that these potential participants are actually qualified for the trials they are interested in?
We don’t ask patients for any information while they’re searching for trials. Once they find a trial they’re potentially interested in, our approach is to use multimodal data enrichment to make sure we are making the right trial recommendations that patients are qualified for.
We need to be able to understand a patient's verifiable medical records and prescription history, do a potential phone screen with the patient using an AI agent, etc., in order to enrich the background. More than 70% of patients consent to us collecting their medical records. Once we have that, we can build a holistic profile for the patient and make highly qualified matches that weren’t possible before, when patients were just filling out a survey online.
How are you working with sites?
We'll work with a sponsor across all the sites on their site list and get all their sites activated on our platform, so they can meet any of the patients in any of the geographies where they may be interested in signing up. As a result, the sites that we work with can be AMCs, professional research networks or independent.
We wanted to build a true co-pilot for sites, versus another portal for sites to manage patient referrals, particularly when sites are already overburdened. To make the process of managing patients easier, one of the things we did is make it easier for sites to see when patients are available for a callback or to book a visit with the site.
"It’s opening the funnel for sites, and it’s helping patients get in touch with people they might not have been in touch with before."
What has been the quantifiable impact you have had on clinical research so far?
I think of our success in terms of what percentage of patient recruitment we represent for a study or for a site, and we represent, on average, 20-40% of the patient recruitment in a study. And when we look at our best sites, the sites that are the most productive on our platform and seeing the most success, we probably represent 20-30% of the patients that they enroll during a given month. One metric from a case study was 1-3 months of recruiting time saved, which is incredibly meaningful from a sponsor’s perspective.
On the patient side, we’ve looked at what percentage of patients were known to sites before we refer them. More than 90% of patients that we end up making referrals for are not patients that the site already had a relationship with. It’s opening the funnel for sites, and it’s helping patients get in touch with people they might not have been in touch with before.
What has been the progression of Power over the course of the year since winning DPHARM Idol Disrupt?
We’re still doing patient recruitment and serving patients, sites and sponsors. However, we’re much bigger, roughly 10 times, than we were this time last year. From a software development standpoint, a lot can change in a year, and the tools that we provide to sites are far more robust, such as in their ability to analyze medical records using AI or to manage patient recruitment activities.
Your company applied AI to help solve a challenge in clinical research. With the rapid acceleration of AI, where is it truly useful and what is still hype?
We use AI every single day in various different ways. We use it in software development and data analysis. We use it especially in understanding user feedback. Every time we get feedback from a research site, sponsor or patient, we use AI analysis to transcribe the notes and summarize the feedback.
It’s “trust, but verify.” I don’t think we can trust AI to make a deterministic call. Anything like AI deciding which patients to screen or enroll, or AI writing an entire protocol, is a bit of a pipe dream and abdicates responsibility for the humans involved. You really need, from an ethical standpoint, an individual making the judgment.
Where AI is very powerful is in doing 80-90% of the work upfront for you. For example, summarizing a patient’s medical record or collecting information from patients. AI can help to shortcut specific laborious tasks. One area I’m excited to apply AI is through the lens of making CRCs’ lives easier and making the lives of sites easier. If we can do 90% of the information gathering – collecting data points, synthesizing and analyzing – that would be a huge win and I think CRCs would be thrilled.
What would be your advice for other entrepreneurs in the digital health space?
The number one piece of advice I always go back to is that it’s never too early to talk to customers, because they will tell you if they are willing to engage or use or pay you for the idea you have in your mind. Often, entrepreneurs can get very quickly wrapped up in an idea without verifying whether there is a significant enough need that would compel someone to use it. If you go to a customer and they aren’t willing, that’s a good signal that the pain point isn’t significant enough.
One of the things we spent time early on working through was which therapeutic areas to focus on. We knew we couldn’t be everything to everyone, especially in the beginning. We used customer feedback to dial in on which therapeutic areas we’re going to say “Yes” to and which ones we would say “No” to for now. Today, our primary focus is on psychiatry, neurology, immunology and cardiovascular/renal.
