Pilot study in Am Samoa explores method to flag colorectal cancer risk using routine blood tests

Pago Pago, AMERICAN SAMOA — (December 10, 2025) The American Samoa Community Cancer Coalition (ASCCC), in collaboration with Lyndon Baines Johnson Tropical Medical Center (LBJ-TMC), the University of North Dakota, and Medical Early Sign, has published the results of a pilot study assessing whether artificial intelligence and machine learning (AI/ML) can help flag individuals at high risk for colorectal cancer (CRC) in low-resource island settings. The study has been published in the Asian Pacific Journal of Cancer Prevention.

Dr. Va’atausili Tofaeono, who led the study team, stated, “For years, our programs concentrated on encouraging individuals to get screened without making much of a change. Over time, we realized the deeper challenges were rooted in the system itself. Instead of setting a 60%, or higher screening target and falling short, we are using an AI/ML model to target those at highest risk for colorectal cancer so we can direct resources where they will have the greatest impact.”

The study evaluated electronic health record data from 6,025 adults aged 50 and older. Using only two common data sources — complete blood count (CBC) results and basic demographic information — the AI/ML model generated CRC risk scores and identified 62 individuals as “high-risk.”

Researchers attempted to reach these high-risk individuals to encourage recommended follow-up testing, including fecal immunochemical testing (FIT), mSEPT9 blood testing, or colonoscopy. Despite these efforts, only four individuals completed any follow-up testing. One participant underwent a colonoscopy, which found a benign polyp and low hemoglobin levels.

This same participant also had an elevated mSEPT9 biomarker, suggesting that pairing AIbased risk scoring with biomarker testing may hold promise in early cancer detection.  The ASCC has already submitted a research proposal that will fully test the efficacy of the AI/ML model.

Colorectal cancer remains the third most prevalent cancer in American Samoa. Barriers to screening include low health literacy, cultural hesitancy, limited access to specialty care, and financial challenges. This pilot study highlights:

 The potential value of AI-assisted risk stratification using data already collected in local clinics

 The critical role of culturally grounded outreach to support follow-up testing

 The need for larger sample sizes and stronger community participation before any AIbased screening method can be implemented

ABOUT THE ORGANIZATIONS

American Samoa Community Cancer Coalition (ASCCC) works to reduce the burden of cancer in American Samoa through screening, education, patient support, and community based research.

LBJ Tropical Medical Center (LBJ-TMC) is the primary healthcare provider in American Samoa and the territory’s central hub for medical services.

Medial EarlySign develops AI solutions that analyze routine clinical data to identify individuals at elevated risk for serious diseases.

The University of North Dakota contributes expertise in public health, epidemiology, and Indigenous health research.