Artificial Intelligence Tool for Optimizing Eligibility Screening for Clinical Trials in a Large Community Cancer Center
Problem
Fewer than 5% of cancer patients enroll in clinical trials and 1 in 5 trials stop for poor accrual, partly because manual eligibility screening from unstructured records is slow.
Why it matters
Poor trial accrual delays evidence generation for new cancer treatments.
What the AI did
An automated matching system used natural language processing to extract patient and trial characteristics from unstructured sources and machine learning to match patients to trials.
Results
Agreement between system and manual attribute extraction ranged from 64.3% to 94.0% across 239 patients and 4 trials; eligibility agreement figure truncated.
Impact
Evidence supports system-assisted eligibility screening with moderate-to-high agreement at one cancer center.
Record
| Venue | JCO Clinical Cancer Informatics |
|---|---|
| Published | 2020 |
| Work type | article |
| Authors | Joseph Thaddeus Beck, Melissa Rammage, Gretchen Purcell Jackson, Anita M. Preininger, Irene Dankwa‐Mullan, M. Christopher Roebuck, Adam Torres, Helen Holtzen et al. |
| Citations | 127 (OpenAlex, at import time) |
| References | 19 |
| Model or method | not stated in source |
| Organization | Highlands Oncology Group (study site) |
| Primary source | https://doi.org/10.1200/cci.19.00079 |
| Open access | closed |
| Keywords | clinical trials, eligibility screening, NLP, machine learning, breast cancer |
Importance rating
significant — thousands to millions affected
Abstract states <5% of cancer patients enroll in trials and 1 in 5 trials stop for poor accrual; slow screening delays evidence generation for new cancer treatments affecting millions.