AI Study Challenges Fears of Racial Bias in Prostate Cancer Care

What You Should Know: 

– Artera, a company developing multimodal artificial intelligence (MMAI)-based prognostic and predictive cancer tests, today announced the publication of a significant validation study in the JCO Clinical Cancer Informatics. 

– The study rigorously assessed Artera’s MMAI model and confirmed its ability to perform similarly across different racial subgroups using data from diverse prostate cancer clinical trials.

Addressing the Challenge of AI Racial Bias in Oncology

This research comes at a crucial time as AI tools are increasingly integrated into clinical decision-making. However, valid concerns persist that these technologies could inadvertently worsen healthcare inequities due to racial bias, particularly if developed using datasets not representative of real-world patient populations. This is a pronounced concern in prostate cancer, where African American men have historically faced a worse prognosis, partly attributed to their underrepresentation in the development of AI tools and genomic biomarkers. Many current AI models have been trained on datasets that lack diversity, raising questions about their applicability and fairness across all patient groups. Artera’s study directly confronts this issue within the context of prostate cancer.

“It’s important to conduct studies to ensure new clinical decision support tools work well across a diverse patient population,” said Mack Roach III, Professor of Radiation Oncology and Urology at the University of California San Francisco and lead investigator for the study. “Especially in the realm of AI, where there can be a lot of healthy skepticism, this study is reassuring to African American men that they too will benefit from using advanced risk stratification tools from Artera.”

Study Confirms Robust Predictive Power Across Diverse Patients

The validation study included a substantial cohort of 5,708 patients sourced from five randomized Phase 3 clinical trials, ensuring a diverse representation. The findings demonstrated that Artera’s MMAI algorithm had a significant ability to predict the likelihood of distant metastasis (DM) and prostate cancer-specific mortality (PCSM) in both African American and non-African American patients with prostate cancer.

These results indicate that Artera’s MMAI model is a critical  tool for identifying patients at higher risk of disease progression or death. This capability allows clinicians to more effectively tailor treatment plans to individual patient needs, irrespective of their racial background. The algorithm’s consistent performance across diverse populations underscores its potential to enhance personalized care for all prostate cancer patients.