Optimizing Clinical Trial Matching using agentic AI

You can also find all our episodes on Spotify if you prefer.

Clinical trial matching is a critical component of advancing healthcare research, yet it remains a significant challenge for many healthcare systems today. Traditional methods of patient identification for clinical trials are often cumbersome, time-consuming, and fraught with inefficiencies. But what if there was a way to streamline this process, ensuring that the right patients are matched to the right trials in real time?

In episode 4 of the Learning Health System podcast, we delve deep into the potential of artificial intelligence (AI) to revolutionize clinical trial matching. If you're a healthcare provider or a professional involved in clinical research, this episode is one you won't want to miss.

The Importance of Clinical Trial Matching

Clinical trials are at the heart of medical advancements. They help to test new treatments, devices, and drugs that could save lives or improve the quality of care for patients. But for a clinical trial to be successful, it’s essential that the right patients are included. This is where clinical trial matching comes in.

Clinical trial matching is the process of identifying eligible patients based on the specific inclusion and exclusion criteria defined by clinical trials. This process can be incredibly complex, as each trial may have numerous, highly detailed criteria that need to be matched precisely to a patient’s medical history, current health status, and other factors.

Traditionally, trial matching has been a manual, labor-intensive process. Healthcare providers, research coordinators, and clinicians must comb through vast amounts of patient data, which can lead to delays, errors, and missed opportunities. In fact, despite the efforts of clinical trial recruiters and coordinators, many patients remain unaware of trials they could benefit from, and many trials fail to recruit enough participants.

With the advent of AI, however, there’s hope for a more efficient and accurate solution.

AI-Powered Solution: (O)TESSA

Enter (O)TESSA, or the (Oncology) Trials Evaluation and Smart Screening Algorithm. This AI-powered tool is designed specifically to enhance the process of clinical trial matching, particularly within oncology, by leveraging advanced machine learning and natural language processing (NLP) capabilities.

Developed by at the Montefiore Health System, (O)TESSA works by analyzing complex clinical trial criteria from sources like clinicaltrials.gov and electronic health records (EHR) systems. It uses large language models (LLama and ClinicalBERT) to break down trial criteria into machine-readable components. From there, the system compares this information to patient data to determine whether a patient meets the trial's criteria.

By automating this process, (O)TESSA not only saves time but also significantly improves the accuracy of trial matching. What might take a human several hours to evaluate can now be done in seconds by (O)TESSA, providing more timely and effective patient recruitment for clinical trials.

How (O)TESSA Enhances Clinical Trial Matching

Traditional trial matching is often riddled with challenges: data overload, inconsistencies in patient records, and the inability to process unstructured data such as clinical notes, imaging reports, and lab results. OTSA is designed to address these issues head-on.

Data Parsing and Normalization

• Data Parsing and Normalization: (O)TESSA breaks down complex, unstructured clinical data into a machine-readable format, enabling faster and more accurate analysis.
• Real-Time Integration with EHR: (O)TESSA integrates seamlessly with existing EHR systems like Epic, enabling “just-in-time” matching.
• Addressing Data Inconsistencies: (O)TESSA ensures that the most reliable and current patient information is used, reducing the risk of errors in trial matching.
• Scalability and Adaptability: (O)TESSA is scalable and adaptable, making it suitable for a wide range of trials beyond oncology.

Improving Fairness and Reducing Bias

One of the most important features of (O)TESSA is its commitment to AI fairness and transparency in trial matching. AI systems, if not carefully designed, can inadvertently perpetuate bias, particularly in areas like clinical trial recruitment, where certain populations are often underrepresented. These biases can skew the outcomes of clinical trials and limit the generalizability of findings.

(O)TESSA is built with this issue in mind. The system incorporates mechanisms to ensure that clinical trials are accessible to diverse patient populations, particularly those who have historically been underrepresented in research. By focusing on equitable patient matching, (O)TESSA ensures that clinical trials offer the potential for groundbreaking treatments to all patients who stand to benefit, regardless of their background, socioeconomic status, or location.

Real-World Impact: (O)TESSA in Action

So, how is (O)TESSA being used in real-world settings? The results so far are promising.

At Montefiore Health System, where (O)TESSA was developed, the AI-powered tool is already being utilized to match patients to oncology trials, specifically focusing on breast and lung cancer. Every day, over 40 trials are reviewed and matched with eligible patients, a process that would have been time-prohibitive with traditional methods. The system is also integrated with EHR systems like Epic, which means that when a patient visits a clinic, the physician can quickly determine if the patient is eligible for any open trials at that time.

The use of OTSA has resulted in faster recruitment, more accurate matches, and, ultimately, better outcomes for patients. But the potential doesn’t stop there. The system is being adapted to work in real-time for critical care settings, such as ICUs, where patients need to be enrolled in trials immediately.

Conclusion

Episode 4 of the Learning Health System podcast provides an in-depth look at (O)TESSA and its role in transforming clinical trial matching. It offers valuable insights into the intersection of AI and healthcare, exploring how machine learning models can be harnessed to improve trial recruitment, reduce bias, and drive better patient outcomes.

For anyone involved in clinical research, IT leadership, or healthcare operations, this episode is a must-listen. The conversation goes beyond just the technology behind (O)TESSA, touching on the broader implications of AI in healthcare and how it can support more effective and equitable patient care.