AI Model Achieves 100% Accuracy in Identifying Prostate Cancer Patients Eligible for Genetic Testing, Study Finds

At the 2026 ASCO Annual Meeting in Chicago, internal medicine resident Dr. Mitchell Singstock unveiled a breakthrough that could change the way men with prostate cancer are screened for inherited risk. His team from the University of Utah reports that a machine‑learning algorithm achieved a perfect 100 % accuracy rate in flagging patients who meet the National Comprehensive Cancer Network (NCCN) criteria for germline and somatic genetic testing. No patient who should have been tested was missed, and no unnecessary test was recommended—an outcome that has long been the holy grail of precision oncology.

The study was retrospective, drawing on a large cohort of prostate cancer cases from the Utah cancer registry. The AI model was trained on thousands of patient records, learning to recognize patterns that align with the NCCN guidelines, which recommend genetic testing for men with metastatic disease, a strong family history of cancer, or those who have undergone specific therapies. When the algorithm was applied to a separate 2025 test set, it matched the guideline‑derived recommendations 100 % of the time. That means every man who should have received a germline or somatic test was correctly identified, and every test that was recommended was truly warranted.

Germline mutations—especially in BRCA1 and BRCA2—are more common in men with prostate cancer than previously thought, and they signal a more aggressive disease. Detecting these mutations offers two major benefits: it can steer clinicians toward targeted treatments—like PARP inhibitors for BRCA carriers—and it opens a window for family members, allowing relatives to screen for inherited risk.

With the AI model validated, the logical next step is to embed it into electronic health record (EHR) systems. When a patient is diagnosed with prostate cancer, the EHR would automatically flag him for genetic testing if the algorithm flags eligibility. A clinician would then receive a clear, evidence‑based recommendation and can order the appropriate germline or somatic test. This automated workflow promises to eliminate the human variability that often leads to under‑testing, especially in busy practices or underserved communities.

The Utah team’s findings fit into a larger narrative of AI’s expanding role in oncology. While algorithms have already improved imaging interpretation and pathology grading, their application to clinical decision support—such as determining who should undergo genetic testing—is still nascent. By showing that a model can match guideline accuracy, the study provides a proof of concept that could be replicated across other cancers and health systems.

Prostate cancer remains the most common non‑cutaneous cancer among U.S. men. The American Cancer Society projects 333,830 new cases in 2026 and 36,320 deaths. NCCN currently recommends germline testing for men with metastatic disease, a strong family history, or those who have received certain treatments. Yet, real‑world testing rates lag far behind guideline recommendations, with many eligible patients never getting a chance to learn their inherited risk.

Automating the identification process could level the playing field by reducing clinician discretion and ensuring every patient meeting NCCN criteria is flagged. This helps close testing disparities, ensuring that patients who might otherwise be overlooked receive timely referrals.

The Utah team plans to pilot the AI‑enabled workflow in a clinical setting. If successful, broader deployment across health systems could standardize genetic testing nationwide, making it easier for patients and clinicians to navigate precision oncology.

In short, the study demonstrates that an AI model can reliably identify men with prostate cancer who should receive NCCN‑recommended genetic testing. By integrating this technology into EHRs, clinicians could streamline referrals, improve access to life‑changing therapies like PARP inhibitors, and help families uncover hereditary cancer risk sooner.