¿Puede la IA diagnosticar endometriosis a partir de irregularidades en el ciclo menstrual detectadas en datos de aplicaciones de seguimiento menstrual ?

La endometriosis interrumpe los ciclos hormonales, a menudo causando patrones de sangrado erráticos. Una IA que analice síntomas registrados en una app podría identificar ciclos atípicos vinculados a la enfermedad. La detección temprana podría reducir los retrasos en el diagnóstico, que actualmente promedian entre 7 y 10 años. La calidad de los datos y los sesgos en los informes de los usuarios siguen siendo obstáculos clave. El enfoque aprovecha patrones de salud obtenidos de multitudes a gran escala.

Background

Endometriosis frequently disrupts menstrual cycles, producing erratic bleeding and symptom records that may differ from typical patterns. A 2023 study demonstrated that machine-learning models analyzing self-reported app data can achieve moderate accuracy in distinguishing probable endometriosis from control groups, yet they still incur high false-positive rates and lack confirmatory imaging or surgical validation—components considered essential for reliable diagnosis.

Because definitive diagnosis currently requires laparoscopic surgery or MRI, AI output based solely on menstrual irregularities is best treated as a preliminary signal rather than a conclusive verdict. Data quality issues, including user-reporting biases and incomplete logs, further complicate the approach. Present systems remain experimental and are not approved for stand-alone diagnostic use; any app-generated alert should prompt consultation with a qualified healthcare provider for appropriate testing.

— Enriched May 12, 2026 · Source: BMJ