New research aims to transform endometriosis diagnosis

In a recent interview with Contemporary OB/GYN, Umair Khanm, graduate student at the University of California (UCSF), San Francisco, and Marina Sirota, PhD, Acting Director at the Bakar Computational Health Sciences Institute at UCSF, discussed current challenges and research advancements in endometriosis screening, diagnosis, and understanding.

The researchers highlighted the primary limitation in diagnosing endometriosis: the significant delay between the onset of symptoms and formal diagnosis, often spanning 5 to 7 years. This delay is largely because of the heterogeneity of symptoms. Some patients may be asymptomatic, while others suffer from severe, debilitating pain. The inconsistent presentation makes it difficult for clinicians to identify the condition early.

To address this, Khan and Sirota conducted a large-scale study leveraging electronic health records (EHRs) from multiple medical centers. Their research aimed to comprehensively examine comorbidities associated with endometriosis. Previous studies had been limited in size or scope, but this broader approach allowed them to identify hundreds of health conditions linked to the disease. These include reproductive, autoimmune, gastrointestinal, neurological, and respiratory disorders. Notably, many of these associated conditions were present even before a formal diagnosis, suggesting a window of opportunity for earlier detection.

The researchers hope their findings will contribute to a better understanding of the various subtypes of endometriosis and inform future diagnostic strategies. One promising avenue is the potential use of artificial intelligence and machine learning to analyze EHR data and predict which patients are at higher risk of developing endometriosis. This could significantly shorten the current diagnostic journey and help clinicians recognize symptom patterns that might otherwise be overlooked.

Sirota and Khan also discussed the potential for integrating richer data sources, such as lab results, medication histories, and even information extracted from clinical notes using advanced language models. The ultimate goal is to develop predictive models that assess an individual’s likelihood of having endometriosis based on their comprehensive clinical profile.

Finally, the researchers emphasized the importance of validating their findings across different health care systems and geographical regions. They also expressed interest in combining clinical data with molecular data, such as biopsy results, to bridge the gap between real-world patient records and deeper biological insights. This integrative approach could pave the way for more personalized and effective care strategies for endometriosis patients in the future.

No relevant disclosures.

Reference

Khan U, Oskotsky TT, Yilmaz BD, et al. Comorbidity analysis and clustering of endometriosis patients using electronic health records. medRxiv. 2025. doi:10.1101/2025.02.13.25322244