Winter 2024 Issue

Cancer Data: How Technology Is Overcoming Challenges in Clinical Trial Matching
By Rajan Gopalakrishnan, MS, and Stephanie Hannigan
For The Record
Vol. 36 No. 1 P. 6

Clinical trials stand as a testament to medical advancement and innovation. They serve as rigorous evaluations of new treatments, drugs, and medical devices, ensuring that they are both safe and effective before they’re introduced to the broader public. Beyond establishing efficacy, clinical trials provide invaluable insights into side effects, optimal dosages, and potential interactions with other treatments.

Behind the scenes, one of the most challenging aspects of these trials is the process of matching patients with suitable clinical trials. This challenge not only slows the advancement of clinical research but also can result in missed opportunities for patients, which can be particularly devastating in instances of cancer or other life-threatening illnesses. However, as the digital age dawns, technology offers hope to revolutionize this space.

In its traditional form, clinical trial matching typically relies on time-consuming manual processes. Clinicians need to sift through extensive clinical trial databases, identifying potential matches based on a patient’s medical history, demographic information, and specific disease attributes. Taking into account the growing workload of today’s clinicians, this manual method of patient screening can sometimes lead to accidental oversights, omissions, and errors. There’s also the issue of trying to keep up with an ever-evolving database of patients and their medical histories, especially in large medical institutions. According to a recent survey, it takes an average of roughly two hours to identify relevant clinical trials for each patient.

At the heart of the clinical trial matching challenge lies the intricacy of trial eligibility criteria. Each clinical trial specifies a set of conditions and exclusions that participants must meet in order to be eligible. These conditions can range from age and gender to specific genetic markers or disease subsets. The complexity of these criteria often makes it difficult for researchers to find suitable participants.

Finding suitable participants quickly is of utmost importance, but only a small percentage of patients get prescreened for clinical trials. Delays in securing participants can push back the trial’s conclusion and, consequently, the potential availability of life-saving treatments to the public. Low enrollment rates can also jeopardize the success of the trial. If there aren’t enough participants, the sample size may be too small for the results to be statistically significant. This poses not only a research challenge but also an economic one, as this can result in wasted resources and funds.

The diversity of potential participants often means that one-size-fits-all research is ineffective. Different populations can respond differently to treatments due to a multitude of factors, including genetics, environment, and socioeconomic conditions. For example, a medication that’s effective in one ethnic group might be less effective, or, in extreme cases, even produce adverse effects, in another. A diverse participant pool is essential for ensuring the safety and efficacy of new drugs and treatments across various populations. By including a representative sample of the population, researchers can better understand and address potential disparities in treatment outcomes.

Enter Technology
This is where technology steps in as a beacon of hope. Imagine a scenario where a patient’s medical record is automatically scanned within the EHR and compared against a comprehensive database of current and upcoming clinical trials, all in real-time and directly at the point of care. In a matter of minutes, the patient and the clinician have a list of potential trials for which the patient matches the eligibility criteria. The clinician, having freed a significant amount of time by not needing to manually search for relevant trials, can then focus time and attention directly on the patient in front of them.

Now, imagine a similar scenario from a different perspective in which clinical research coordinators have access to an automatically updated list of open clinical trials. Using data extracted from the EHR and other health records, they would be presented with a list of the most qualified patients for a particular clinical trial, each of which is assigned a match score based on the patient records’ alignment with the selected clinical trial’s inclusion and exclusion criteria. The clinical research coordinator can then quickly and efficiently identify patients who are most suitable for the clinical trial.

Innovations like artificial intelligence (AI) and natural language processing (NLP) are making scenarios like these an attainable reality. Patients can gain access to treatments and interventions that may have otherwise been difficult to obtain, and clinical trials can achieve higher enrollment rates, ensuring that medical advancements reach the public in a shorter amount of time than traditional manual methods would allow.

The term AI covers a range of technologies and techniques that emulate human-level cognitive capabilities. In health care, AI is already being applied successfully to enhance clinical decision making, automate administrative functions, and support research. NLP, a subset of AI, can read and understand the context within vast text-based data sources like EHRs, pathology reports, radiology reports, genomic reports, and even physician notes.

Medical records contain complex, unstructured data that are often written in very variable nonintuitive medical terminology and acronyms spread across disparate documents. By using NLP, researchers can quickly identify patients whose medical histories align with a trial’s eligibility and exclusion criteria. This extraction process ensures that all potential trial matches are considered, even if they are buried deep within these dense medical documents. But it’s not just about extracting data. NLP tools can understand the context behind medical terminology, ensuring that matches are contextually relevant.

AI and NLP can identify potential matches with accuracy and precision, significantly speeding the matching process. What might take a human several hours, AI and NLP can accomplish in minutes. By automating these time-consuming manual tasks, clinicians are free to invest more time in direct patient care, research, and other high-value activities. Not only does the combination of these two technologies offer time savings but it also reduces the margin of human error.

While these technologies offer a transformative approach to clinical trial matching, their widespread adoption faces barriers. There’s a general distrust of AI-generated suggestions based on concerns around explainability and fairness. For many decision-makers and users, these technologies are black boxes that have not proven themselves in all the different scenarios. There are also fears about whether these tools will replace human expertise.

However, it’s essential to view these tools as augmentative rather than replacements for human expertise. AI and NLP can handle the bulk of data sifting, but human experts still need to review and verify these automated recommendations. This combined approach ensures both speed and accuracy.

As with any digital transformation, there’s also the challenge of integrating new systems into existing infrastructures. Effective integration of clinical trial matching tools and technologies requires a careful assessment of where they fit within an institutional landscape. A good deployment also calls for user workflow analysis as well as oversight. Finally, there needs to be a framework in place to measure the outcomes and effectiveness of these new systems.

The world as we know it continues to evolve, and it’s clear that the advancement of modern-day clinical trials depends on harnessing the power of technology. The capabilities of AI and NLP not only overcome the limitations of traditional methods but also open doors to possibilities we’ve not yet fully realized. The fusion of technology and health care promises a brighter, more efficient future for clinical research and the countless lives it affects.

— Rajan Gopalakrishnan, MS, is the director of informatics and IT at the University of Chicago Medicine Comprehensive Cancer Center. With more than 26 years of experience in technology and 16 years in health informatics, he specializes in digital transformations of clinical research and public health enterprises. In addition, Gopalakrishnan is a Cancer Informatics Advisory Group member at North American Association of Central Cancer Registries.

— Stephanie Hannigan is a marketing manager at Inspirata, a leading oncology informatics company offering several innovative software solutions leveraging scalable artificial intelligence and natural language processing in the clinical trial and cancer registry spaces. Having worked for Inspirata for eight years, Hannigan has developed a strong passion for the application of smart technology in automating oncology informatics. To learn more, visit www.inspirata.com.