Background

In the history of medical research, women have been systematically excluded for a long time. Randomized Controlled Trials (RCTs), which have served as the standard for drug development for decades, were designed primarily around male subjects under the pretext of minimizing variables such as hormonal cycle variability or pregnancy potential. Although female participation rates have gradually increased since the enactment of the NIH Revitalization Act in 1993, the actual accumulation of clinical evidence still lags behind male-centric data.

Gender-based differences in drug responses observed in cardiovascular disease, autoimmune diseases, and specific oncology fields remain unresolved risk factors in clinical practice. Even though myocardial infarction symptoms manifest differently in women than in men, or the blood clearance rate of certain drugs is significantly lower in women, appropriate dosage guidelines are often recommended based on male standard body types. Female-specific diseases such as endometriosis, polycystic ovary syndrome (PCOS), and peri-menopausal metabolic disorders continue to be deprioritized in basic pathophysiological research and prospective clinical trial support.

Traditional RCTs, conducted in strictly controlled environments, struggle to timely address the growing demand for women's healthcare and fill the accumulated knowledge gap. The structure of clinical trials, which requires years for patient recruitment and massive budgets, is structurally limited in rapidly conducting studies on complex chronic diseases or long-term follow-up research. This is the background for the urgent need for a new paradigm of evidence generation.

Key Findings

This analysis published in Nature Medicine focuses on the combination of Real-World Data (RWD) and citizen science as an alternative to overcome the speed limitations of traditional trials. The researchers analyzed that Electronic Health Records (EHRs) generated in routine clinical settings, National Health Insurance claim data, and spontaneous reporting systems for adverse drug reactions are effective in capturing treatment patterns among female patients. Data from wearable device sensors actively recorded by individuals outside the hospital, as well as data from menstrual cycle tracking mobile applications, were also evaluated as precise indicators for monitoring disease prognosis.

The contribution of the citizen science model, which transforms patients from passive subjects into active participants in research design and data collection, was also highlighted. This is because it allows for the rapid aggregation of patient-reported outcomes regarding subtle symptom changes and quality-of-life indicators that have been overlooked by existing clinical systems. As typical examples, patient-led self-reporting registries in areas with high female incidence and unclear standard biomarkers, such as endometriosis or Long COVID, have served as a foundation for revealing symptom manifestation patterns and drug adherence.

Researchers presented comparative cases demonstrating how RWD analysis was used to validate gaps in existing RCT data. Post-hoc analysis of hundreds of thousands of health insurance records confirmed that certain analgesics and antidepressants, approved based on male-biased samples, cause more frequent adverse reactions in women. Quantitative achievements, such as completing patient monitoring—which typically took over five years in traditional clinical trials—in just a few months using decentralized registries and crowdsourced remote data collection tools, also support the validity of the alternative data model.

Implications and Prospects

This proposal calls for a comprehensive overhaul of data integration strategies by pharmaceutical companies and regulatory authorities. It aligns with the trend of the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA) specifying guidelines on the use of Real-World Evidence (RWE) for post-marketing safety assessments and approval of expanded indications. Gender-stratified analysis must be mandated from the clinical design phase, and a hybrid trial design incorporating real-world data (RWD) should be actively adopted to ensure the effectiveness of clinical practice guidelines.

Standardizing citizen science-based data and establishing data governance remain challenges to be solved. Subjective symptom records entered by individuals via smartphone apps are prone to measurement bias, and the rate of data omission is higher than in hospital-based clinical trials. Establishing de-identification technologies and data protection protocols to prevent the commercial misuse of sensitive reproductive health information and location-based sensor information is also a prerequisite. The construction of machine learning-based quality verification pipelines to ensure data reliability is expected to become more active.