BehavioralPulse™: Enhancing Trial Site Prioritization

Product: BehavioralPulse^TM 

Industry & Function: Life Sciences - Clinical Trials

Therapeutic Area: Oncology, Neuroscience, Immunology

Summary

Surgo Health’s BehavioralPulse data was used to improve site enrollment prediction in a large retrospective study with a top pharma client. The sponsor’s existing feasibility model, built using clinical and social determinants of health (SDOH) data, often identified sites as high-potential that ultimately struggled or failed to recruit patients. When added to the sponsor’s model, Surgo’s novel behavioral data improved site rankings across 290 phase 2 & 3 studies in oncology, neuroscience, and immunology.

What Traditional Feasibility Models Miss, and Why It Matters for Clinical Trials

Industry-standard feasibility models rely on historical site performance, clinical volume, and social determinants of health to predict recruitment success. While these factors are valuable, they often miss the behavioral and contextual factors that influence whether eligible patients actually enroll. By enriching existing models with behavioral data, BehavioralPulse helps organizations:

• Improve the accuracy of site selection and enrollment forecasting
• Identify high-performing sites that traditional models overlook
• Prioritize sites more likely to recruit representative patient populations
• Reduce costly investments in underperforming trial sites
• Make more informed site selection and resource allocation decisions

This missing behavioral intelligence enhances traditional feasibility models by adding the missing behavioral context behind patient participation, enabling more accurate site prioritization and better clinical trial outcomes.

Challenge 

Pharmaceutical sponsors heavily rely on site feasibility models, leveraging historical site performance, clinical volume data, and SDOH variables to identify high-performing trial sites. Yet, predicting site performance remains difficult and inefficient, leading to large investments in clinical trial sites that are unable to successfully enroll patients.

Approach

Surgo Health performed a retrospective analysis of 290 Phase 2 & 3 studies with 8,602 study-sites that had recruitment performance data. The studies covered multiple myeloma (MM), major depressive disorder (MDD), and inflammatory bowel disease (IBD).

The sponsor’s existing feasibility model predicted site performance using industry-standard variables: historical performance (past enrollment rates of the investigator/site), clinical volume (real world data, prevalence of the indication ICD-10 codes) in the catchment area, and standard SDOH (e.g., income, education).

Surgo appended 70+ proprietary BehavioralPulse features to the catchment area of each site, including non-clinical variables including:

• Medical distrust
• Transportation insecurity
• Health literacy
• Belief in race-equity in research

Performance of the control model (clinical and SDOH data) was compared against the enriched model (control plus BehavioralPulse data). Feature importance was also calculated using SHAP values (SHapley Additive exPlanations), quantifying the improvement added by each behavioral variable.

Results 

BehavioralPulse features improved the accuracy of predictive models by 26% across indications (see Table).

Enriched models additionally improved prediction of representative recruitment rates across sites in IBD and MM, particularly for identifying sites with Black participant enrollment in the bottom quintile (25% improvement).

SHAP analysis revealed that 8 out of the top 20 most predictive features for site success were behavioral or contextual (Surgo data).

The improvement informed by behavioral data demonstrates that traditional feasibility models struggle to differentiate between sites with similar clinical volumes and SDOH, but vastly different patient behavioral factors.

Table: Baseline and enhanced model performance comparison for overall recruitment. The score represents the Spearman rank correlation, where higher values indicate better model performance.

See How BehavioralPulse Strengthens Trial Planning

Improve site selection and enrollment forecasting with behavioral data that complements traditional feasibility models and enables smarter clinical trial decisions.

Talk to an expert about applying BehavioralPulse to your next clinical trial.