AI-Assisted HPV Screening Refines Cervical Precancer Risk Across Nine Countries
A 50,000-participant study found that HPV genotyping and AI-assisted cervical imaging supplied complementary signals for triaging severe precancer in resource-limited settings.
Quick summary
- A study across nine lower-resource settings combined self-collected HPV testing, extended viral genotyping and AI-assisted analysis of cervical images.
- More than 50,000 women aged 25 to 49 were enrolled, and more than 500 cases of severe precancer were identified among participants with valid results.
- The approach may help health systems prioritize follow-up and support same-day screening, triage and treatment, but it is not yet proof of population-level impact or autonomous diagnosis.
A large international study suggests that artificial intelligence can add useful risk information to human papillomavirus screening in places where specialist capacity is limited. The PAVE consortium evaluated a pathway that starts with self-collected HPV samples, adds extended genotyping for 13 cancer-causing HPV types, and then uses automated visual evaluation to analyze cervical images from women who test positive.
The cross-sectional study enrolled more than 50,000 women aged 25 to 49 between 2022 and 2025 in Brazil, Cambodia, the Dominican Republic, El Salvador, Eswatini, Honduras, Malawi, Nigeria and Tanzania. Participants with a positive HPV test received imaging with a portable colposcope and a biopsy for diagnostic confirmation. A deep-learning system classified the images as severe, indeterminate or normal.
The main finding is that HPV type and the AI-assisted image assessment supplied complementary information about the risk of severe precancer, known as CIN3+.
Among nearly 49,000 participants with valid results, more than one quarter tested positive for HPV, according to the research team. The study identified more than 500 CIN3+ cases, with a higher prevalence among women living with HIV. HPV16 carried the highest risk, particularly when the automated image assessment also found severe abnormalities.
Why the combination matters
HPV testing is sensitive, but a positive result does not by itself show who has a high-grade lesion and needs rapid treatment. In resource-constrained health systems, sending every HPV-positive patient through the same follow-up pathway can overwhelm available staff and equipment. A portable imaging step could help separate higher-risk cases from those that can be managed less urgently.
The design also supports a possible single-visit model. Self-sampling, rapid HPV testing and real-time image analysis could reduce the number of return appointments, which is important where travel, cost and limited clinic capacity make follow-up difficult.
Reality check
This was a cross-sectional evaluation, not a randomized trial of screening programs. It shows that the combined signals predicted biopsy-confirmed precancer in the study population, but it does not establish how much the strategy reduces cancer incidence or mortality, how often performance changes between clinics, or whether local health systems can operate it safely at scale. The AI output is a triage aid within a clinical pathway, not a replacement for trained staff, biopsy confirmation or quality assurance.
What comes next
The strongest next evidence would come from prospective implementation studies that measure missed lesions, unnecessary treatment, patient follow-up, cost, device maintenance and performance across different populations. If those results hold, the approach could make cervical-cancer screening more accessible while directing scarce clinical resources toward women at greatest risk.
Verified topics and entities
Sources and citations3 sources
External references used to support the reporting in this article.
Published by
NewTqnia Health Desk
An institutional editorial team within NewTqnia
