Life Sciences Companies Are Using AI to Recruit Oncology Patients Faster

Life Sciences Companies Are Using AI to Recruit Oncology Patients Faster

Patient recruitment has always been one of the major challenges in oncology clinical research, with many trials failing to find and recruit the right patients. With AI, we can now contextualize patient journeys from unruly electronic medical record (EMR) data with unprecedented speed and precision.  

When recruiting for a trial, research staff must compare patient data against the trial’s inclusion and exclusion (I/E) criteria to find potential matches. However, more than 80% of patient data stored in EMRs is unstructured, such as physician notes, imaging, lab reports, family history, etc. Using traditional methods, this unstructured data is much more tedious and time-consuming to sort through versus coded, structured data. There are many medical terms that don’t have a standardized code and may be documented differently depending on the clinician or healthcare organization (HCO). For example, muscle-invasive bladder cancer (MIBC) does not have a specific code and may appear in notes as ‘muscle invasive bladder cancer,’ ‘MIBC,’ ‘advanced bladder cancer,’ ‘invasive bladder cancer,’ ‘high grade bladder cancer,’ ‘advanced urothelial bladder cancer,’ etc. We did a test in the Deep 6 AI software to see how precisely we can find patients with MIBC. First, we searched MIMIC EMR data at a single health system by code only. We found 40,000 patients with ‘bladder cancer’ using the ICD-10 code C67. Then, we applied AI and mined clinician notes and disparate lab reports to find patients with ‘muscle-invasive bladder cancer.’ AI-assisted chart review identified 10,000 patients with MIBC, pinpointing MIBC four times more precisely than by searching coded EMR data.

Not only does it improve precision, AI also improves matching and validation speed. Manual chart review typically takes about four hours of work to identify one patient. Research staff need to manually review the charts of four patients with bladder cancer to find the one patient with MIBC. Using AI-assisted review, it takes less than two minutes to find a patient with MIBC. Therefore, validating patients is 100 times faster using AI. 

Clinical stage cellular immuno-oncology company accelerates recruitment for a non-small cell lung cancer trial

AI has had a tangible impact optimizing study design, improving feasibility, and accelerating recruitment at clinical research sites. One HCO was recruiting for a clinical stage cellular immuno-oncology non-small cell lung cancer trial. Prior to using the Deep 6 AI software, they were manually reviewing doctor’s notes in the EMR which resulted in recruiting zero patients in five months and taking each staff member eight hours per week to screen patients. Using Deep 6 AI, they enrolled 124 patients in five months and reduced the time spent screening to two hours a week. Screening four times faster saved the HCO $52K in less than one year.

Pharmaceutical company accelerated recruitment by 33% for a colorectal cancer clinical trial

The pharmaceutical company was recruiting patients to a clinical trial for a second-line therapy for colorectal cancer. Due to complex I/E criteria, including failed first-line therapy and specific genetic mutations (KRAS-mutated mCRC), it was difficult to find the right participants and multiple sites were having trouble meeting enrollment goals of one patient per site every three months. By tapping into the Deep 6 AI ecosystem, the company running the trial was able to query EMR data to find precise cohorts of eligible patients and share the cohorts with site staff for validation. In less than two months with Deep 6 AI, the company was able to increase their enrollment speed by 33% across two sites. Prior to completing enrollment, the FDA advanced the therapy from second-line to first-line.

NCI-designated Comprehensive Cancer Center reduces average recruitment duration by seven months using AI

A premier oncology academic medical center (AMC) evaluated 40 studies to compare recruitment progress before and after AI was used to identify patients. They found that recruitment was streamlined after shifting away from the manual chart review process. Using Deep 6 AI, they were able to demonstrably reduce the gap in their recruitment goals and reduce the time spent recruiting for a trial by seven months, on average.

Through these real-world examples, we can see how AI is precisely matching patients to complex I/E criteria and surfacing hard-to-recruit patient populations. Learn more about the Deep 6 AI Recruitment Acceleration solution here

About The Author

Nour Malki