Clinical trial matching: the frustration of matching patients to cancer research begins with data issues — and can end with artificial intelligence
If you work in oncology or cancer clinical trial coordination in the US, you already know the numbers don’t always work in your or your patients’ favor.
Less than 5 percent of adult patients with cancer participate in clinical trials,1 and one in 10 cancer trials registered on ClinicalTrials.gov between 2005 and 2011 were closed prematurely due to poor accrual and other trials taking longer than anticipated to complete recruitment.2
In a healthcare system that focuses on innovation and continuous improvement, how can there be such a disconnect in clinical trial recruitment?
Manual processes, unstructured data and imprecise criteria
The problem typically lies in finding matches within available data — identifying the right patients with the right clinical attributes who meet highly specific eligibility criteria for a relevant clinical trial.
That data-matching identification process can be daunting – and is still performed manually at most medical centers. Staff are charged with reviewing complex and lengthy electronic health records (EHRs), and creating multifaceted spreadsheets to guide the matchmaking.
Complicating matters is that about 80 percent of healthcare data is unstructured,3 making traditional rule-based alignment between patients and trials difficult. Plus, available trial criteria can be vague or incomplete.
How to solve the problem of cumbersome clinical trial identification? Artificial intelligence.
The evolution of artificial intelligence (AI)-based screening processes is being driven by the pressing need to automate data-driven trial identification steps and increase the speed at which matches can be made.
An AI-based solution is able to process and link large amounts of both structured and unstructured patient EHR data, medical literature and trial information, and eligibility criteria from clinicaltrials.gov.
The use of natural language processing allows an AI-based system to analyze all of that collected data. And AI can take unstructured, inexact and lengthy criteria lists and learn how to interpret the trial requirements based on every patient case.
So, without any manual intervention, an AI-based solution can understand both the clinical data in the medical record, as well as the inclusion and exclusion criteria of the trials, and it can do it quickly. In one IBM Watson Health™ study, AI-based identification cut the time required to screen patients for clinical trial eligibility by 78 percent.4
In minutes, staff can access a list of trials for which a patient is potentially eligible, and equally as important, a list of trials from which the patient has been excluded based on eligibility.
From this detailed analysis, clinicians can decide which trials to explore further as treatment options. Using the same analysis, research trial coordinators are able to continuously review their patients for potential eligibility to meet their recruitment targets.
The future of clinical trial recruitment belongs to AI
As the adoption of AI technology and clinical trial matching services increase in clinical trial recruitment — particularly in academic and research medical centers around the nation — patient screening efficiency and capacity will only increase, and more effective trial recruitment will become the norm.
That’s good news for patients, healthcare providers and science.
- “Role of Clinical Trial Participation in Cancer Research: Barriers, Evidence, and Strategies“, Joseph M. Unger, Ph.D.,1 Elise Cook, M.D.,2 Eric Tai, M.D.,3, Archie Bleyer, M.D.4, July 2017, ncbi.nlm.nih.gov.
- Moorcraft SY, Marriott C, Peckitt C, Cunningham D, Chau I, Starling N, Watkins D, and Rao S. Patients’ willingness to participate in clinical trials and their views on aspects of cancer research: results of a prospective patient survey. Trials, 2016. https://www.ncbi.nlm.nih.gov/
- “Data-driven healthcare organizations use big data analytics for big gains,” IBM white paper, February 2013. https://www.ibmbigdatahub.com/
- Beck J, Vinegra M, Dankwa-Mullan I, Torres A, Simmons C, Holtzen H, Urman A, Roper N, Norden A, Rammage M, Hancock S, Lim K, Rao P, Coverdill S, Roberts L, Williamson P, Howell M, Chau Q, Culver K, Sweetman R. Cognitive technology addressing optimal cancer clinical trial matching and protocol feasibility in a community cancer practice. J Clin Oncol. 2017;35 (suppl; abstr 6501). doi: 10.1200/JCO.2017.35.15_suppl.6501. https://ascopubs.org/