top of page

Reducing Sample Size While Improving Equity in Vaccine Clinical Trials: A Machine Learning-Based Recruitment Methodology with Application to Improving Trials of Hepatitis C Virus Vaccines in People Who Inject Drugs

AI in Medicine

March 2024

[Full Text] [Pubmed]


Background and Aims A vaccine for Hepatitis C virus (HCV) could greatly reduce the burden of HCV, particularly in high-risk groups such as people who inject drugs (PWID). However, recruitment to such a trial must overcome the twin challenges of high trial costs and disparities in enrollment. Here we investigate trial recruitment informed by machine learning and evaluate a strategy for HCV vaccine trials termed PREDICTEE - Predictive Recruitment and Enrichment method balancing Demographics and Incidence for Clinical Trial Equity and Efficiency.

Methods PREDICTEE utilizes a predictive model applied to trial candidates, considering their demographic, network, and injection characteristics collected via questionnaire. Using these questionnaire responses, the model outputs a probability the candidate will be infected with HCV during the trial. The decision to recruit considers both the candidate’s predicted incidence and the vaccine’s target population. We evaluated PREDICTEE via in silico methods, in which we first generated a synthetic candidate pool using a model of PWID in Chicago. We then compared PREDICTEE to conventional recruitment of high-risk PWID who share drugs or injection equipment in terms of sample size and participation-to-prevalence ratio across age, sex and race.

Results Comparing conventional recruitment to PREDICTEE found a reduction in sample size from 802 (95%: 642–1010) to 272 (95%: 264–288) with PREDICTEE, while also reducing screening requirements by 31%. Simultaneously, participation-to-prevalence ratio increased from .475 (95%: .356–.568) to .751 (95%: .664–.841). Even when adjusting to a highly dissimilar maximally-balanced target population, PREDICTEE is still able to reduce required sample size from 802 (95% 642–1010) to 300 (95%: 288–314) while improving participation prevalence ratio to .809 (95%: .794–.827).

Conclusions PREDICTEE presents a promising strategy for HCV clinical trial recruitment, achieving sample size reduction while ensuring equitable recruitment in a diverse set of scenarios and target populations.

Partnerships that get results: Global. Fluid. Connected. Join Us!

Web Footer_LOGOS_Jan 2023_rev.jpg

Loyola University Medical Center

Department of Medicine

Division of Hepatology

2160 S. First Ave
Mulcahy Center, Rm 1610

Maywood, IL 60153, USA


Phone: 708-216-4682

Fax: 708-216-6299

bottom of page