Can AI diagnose endometriosis from menstrual cycle irregularities detected in period-tracking app data ?

Endometriosis disrupts hormonal cycles, often causing erratic bleeding patterns. AI analyzing app-recorded symptoms could identify atypical cycles linked to disease. Early detection could reduce delays in diagnosis, which currently average 7–10 years. Data quality and user-reporting biases remain key hurdles. The approach leverages crowdsourced health patterns at scale.

Researchers have begun exploring whether AI models can use symptom logs from period-tracking apps to flag possible endometriosis, but current systems remain experimental and are not approved for diagnosis. A 2023 study showed that machine learning on self-reported symptoms can achieve moderate accuracy in distinguishing likely endometriosis from controls, yet high false-positive rates and the lack of confirmatory imaging or surgical validation limit clinical reliability. Because endometriosis diagnosis typically requires laparoscopic surgery or MRI, AI suggestions based only on cycle irregularities should be viewed as preliminary signals rather than definitive diagnoses. Any alerts from an app should prompt consultation with a qualified healthcare provider for appropriate testing.

— Enriched May 12, 2026 · Source: BMJ