Using a Multilingual AI Care Agent to Reduce Disparities in Colorectal Cancer Screening: Higher FIT Test Adoption Among Spanish-Speaking Patients

Meenesh Bhimani,R. Hal Baker,Markel Sanz Ausin,Gerald Meixiong,Rae Lasko,Mariska Raglow-Defranco,Alex Miller,Subhabrata Mukherjee,Saad Godil,Anderson Cook,Jonathan D. Agnew,Ashish Atreja
DOI: https://doi.org/10.1101/2024.12.16.24318586
2024-12-18
Abstract:Background: Colorectal cancer (CRC) screening rates remain disproportionately low among Hispanic and Latino populations. While artificial intelligence (AI) has shown promise in healthcare delivery, its impact on health equity remains unclear. Objective: To evaluate the effectiveness of a bilingual generative AI voice agent outreach program in engaging Spanish-speaking patients for CRC screening compared to English-speaking patients. Methods: We conducted a retrospective analysis of AI-powered outreach calls for CRC screening at a large integrated health system serving central Pennsylvania and northern Maryland in September 2024. The study included 1,878 patients (517 Spanish-speaking, 1,361 English-speaking) eligible for colorectal cancer screening. The AI care agent conducted personalized phone calls in the patient's preferred language to discuss screening and facilitate fecal immunochemical test (FIT) kit requests. The primary outcome was FIT test opt-in rate. Secondary outcomes included call connect rates and duration. Results: Spanish-speaking patients demonstrated significantly higher engagement across all measures compared to English-speaking patients: FIT test opt-in rates (18.2% vs. 7.1%, p<0.001), connect rates (88.8% vs. 53.3%, p<0.001), and call duration (6.05 vs. 4.03 minutes, p<0.001). In multivariate analysis, Spanish language preference remained an independent predictor of FIT test opt-in (adjusted OR 2.012, 95% CI 1.340-3.019, p<0.001) after controlling for demographic and other factors (gender, age, state of residence, and call duration). Conclusions: Contrary to concerns about technology exacerbating disparities, AI-powered outreach achieved significantly higher engagement among Spanish-speaking patients. These findings suggest that language-concordant AI interactions may help address healthcare disparities and improve preventive care engagement in traditionally underserved populations.
What problem does this paper attempt to address?