AI Speeds Patient Recruitment At Health Quest
By Ed Miseta, Chief Editor, Clinical Leader
Artificial intelligence (AI) has the potential to disrupt and revolutionize many aspects of clinical trials. One area in which it is already making an impact is patient recruitment. Mining electronic medical records (EMRs) for patients meeting inclusion and exclusion criteria can be tedious and time-consuming work but is a chore that seems ideal for an AI solution.
Health Quest is a four-hospital nonprofit system operating in New York’s Mid-Hudson Valley and northwestern Connecticut. The group recently signed an agreement with IBM Watson Health to start matching its patients to clinical trials for which they may be eligible.
Health Quest began working with IBM Watson in early 2019 on a program to help with population health management. Health Quest needed access to real-time patient data to meet quality-based performance benchmarks and was able to use Watson Health to consolidate disparate data from across practices and departments. The result of that effort was the closure of gaps in care and the generation of $3.7 million in billing revenue.
Although that may sound like a rudimentary task, Loomis states it is not. The old model was difficult and time consuming and required Health Quest to pay nurses to spend hours going through medical records and trial information, trying to determine which patients were the best fit for trials that were recruiting. Even if the nurses were successful in that endeavor, they were only qualifying patients for the trials they were aware of. There were many trials, including some that would be a better fit for the patient, that the nurses might simply be unaware of.
“The clinical trial software will allow us to qualify more patients for more trials, and do it in a much timelier manner,” says Loomis. “We will also be matching them to the best possible trial for their particular cancer and for their particular circumstance. Some of those trials will be within our own organization and some may be trials that we are not offering. Regardless, we are looking for the trial that will provide each patient with the best possible treatment available and give them the best chance for an optimal outcome.”
Precision Medicine Complicates Recruitment
The growth of precision medicine treatments has certainly made clinical trials more complex. This has also put additional inclusion and exclusion criteria in place for patients. While the newer treatments being developed present a greater likelihood of success for patients, it also creates additional work for those attempting to locate patients.
Loomis notes this is a big part of the reason why hospitals require a lot of individuals working a lot of hours to comb through medical records in order to locate the right patients for a study. Even when you locate qualifying patients, there is no way to know if you are connecting them with the most optimal trial for their situation. What makes them the perfect candidate for a trial is dependent upon their individual situation, their type of cancer, and other conditions they may be dealing with. It is very difficult, if not impossible, for a human to be able to review and digest all that information.
The AI solution, called Watson for Clinical Trial Matching, is a cognitive computing system programmed to match patients to trials for which they may be eligible. Information on patients can be input, along with a complete catalog of all available trials and their inclusion/exclusion criteria. The system will then report the best available trial for that patient. That trial could be within Health Quest, another local hospital, or a large cancer center Such as MD Anderson in Houston. Clinical Trial matching has already been shown to cut patient screening time by 78 percent. In breast cancer trials conducted at the Mayo Clinic, it drove an 84 percent increase in enrollment in the first 18 months after implementation.
Health Quest always attempts to get patient consent upfront. Patients are asked in person or via email if they have an interest in participating in a clinical trial if one can be matched to their condition. Once the consent is received their records can be examined and matched to potential trials.
“At one time, many clinical trials were only available to patients at large hospitals and healthcare centers in major urban areas,” says Loomis. “We are very excited about the ability to now offer this same level of care to patients in the mid-Hudson Valley.”
Solve The Patient Recruitment Bottleneck
Patient recruitment remains a problem for researchers, but Loomis believes AI and efforts like the one underway at Health Quest will be a big part of the solution. A lot of time and money are spent trying to find trials that are the best match for a patient. Doing so in a more automated fashion makes that task far easier for both clinicians and patients.
“AI is helping us match patients to the right trial, but it will also make it much easier for us to get patients interested and involved in trials,” adds Loomis. ‘We know we have patients who would like to be involved with a trial, but for whatever reason are unable to locate the right trial. At times we also did not have enough resources to devote to helping patients find the right trial. AI will make that process easier for us and allow us to bring more trials to more patients.”
AI is part of a broader strategy for the Health Quest system. Loomis sees AI as a tool that will become a large disruption to both medicine and clinical trials. Health Quest is now making an active push to incorporate the technology into many aspects of its system to make the network more efficient and effective. Clinical trial matching is seen at this time as just the starting point for that strategy.
Loomis points out the fragmented nature of electronic health records as one area where AI can create efficiencies. He notes the technology can scan those records and provide information to clinicians at the point of care. He believes that is the area where AI has the potential to really shine. For example, a natural language processing engine scanning through all notes on a patient and providing insights to that information in one location. AI is also able to take that information, compare it to databases, and bring back recommendations for best practices.
“What we are doing now is a very early use case for AI,” says Loomis. “We are taking patient information and comparing it to trial information. I see this as a very early indicator of things we will soon be able to do in a much bigger way.”