Diagnosis Female: How Data Bias Endangers Women’s Health
Allison Teti, University of Pennsylvania
Too often, women’s experiences are being overlooked or discounted by doctors and healthcare professionals. This leads to a trust-gap that negatively affects women’s health. There is a widening body of research focusing on the clinical implications of gender differences in the healthcare setting. Medical knowledge is generally taught to be gender neutral to medical and health professionals, resulting in a lack of gender competency in healthcare. This incompetency contributes to the deep-rooted methodical problems that predicate women’s experiences of misdiagnosis and feeling dismissed by the medical system (Cleghorn, 2021).
The history of medicine is just as much social and cultural as it is scientific. Medical fallacies about gender dating back to ancient Greece continue to negatively impact medical treatment of women today. The practice of medicine has absorbed and enforced the socially constructed gender divisions that have generally ascribed authority, power, and dominance to men. Patriarchal culture and the superiority of the male body has been cemented into medicine’s foundational premises. For example, Aristotle, commonly regarded as the first great biologist, contended that the female body was a result of a male whose development ended too early in the womb, presenting women as anatomically inferior to men. The only caveat to this system of thought was that females possessed “an organ of the highest biological — and social — value”, the uterus. Ultimately, having this organ contributed to the idea that a woman’s worth was determined by her ability — and responsibility — to carry and rear children. Women’s illnesses were consistently related back to the “secrets and oddities of her reproductive organs,” which led to a constrained definition of what it meant to be a woman (Cleghorn, 2021).
Despite advances in medical understanding of female biology, medicine has continues to conflate biological sex with gender identity, reflecting and reinforcing socio-cultural assumptions about women. As a result, when women do not identify with the gender standards assigned to biological femininity by medicine, their access to health care is limited. These barriers become amplified for women of color/ Specifically, women who identify as Black, Asian, Indigenous, Latinx, and other women of color face increased discrimination and bias in the healthcare system (Cleghorn, 2021).
Social movements and advocacy efforts have had a significant impact on advancing health research. For example, the late 20th century women’s health movement, which frequently relied on lived expertise to highlight gaps and injustices, is responsible for drawing attention to sexist medical practices, the exclusion of females and women in clinical trials, conformity to the “male norm” in medicine, and underresearched women’s conditions and bodies (Greaves & Ritz, 2022). It is alarming to note that the NIH inclusion policy was not codified into Federal law until 1993, when Congress passed the NIH Revitalization Act of 1993 (Public Law 103–43) and included a section titled Women and Minorities as Subjects in Clinical Research. The resulting gender data gaps put women at risk in many ways such as a disproportionate amount of care delays, diagnosis errors, higher rates of adverse drug reactions, and lower survival rates. For example, endometriosis, a gynecological disease that affects one in ten women, typically takes four to ten years to diagnose. This is partially a result of many physicians dismissing or doubting female patients’ reports of severe or chronic pain and partially due to gaps in research about the disease (Cleghorn, 2021; Wood, n.d.).
Currently, gender bias in healthcare is receiving greater public attention, and more women are attempting to raise awareness of the ubiquitous and pervasive issue by sharing their own experiences. Eye opening exposés such as My Doctor Didn’t Believe My Pain and the Netflix Documentary “The Bleeding Edge’’ chronicle the devastating experiences of women who have been dismissed, minimized, and/or misdiagnosed as a result of medical sexism. However, gender-based medical fallacies were developed long before medicine became an evidence-based discipline; these prejudices continue to have a negative influence on the standard of care, treatment, and diagnosis for all persons who identify as women (Cleghorn, 2021).
Furthermore, the medical industry is rapidly employing machine learning algorithms to direct patient care and administrative activities, which poses a severe risk to the field’s future. These algorithms are being used in various applications, such as to detect disease from radiology images, guiding clinical trial designs, and recommending personalized treatment plans for patients. Because machine learning algorithms work by learning from the past — using historical data to deduce statistical patterns — they absorb the biases and assumptions embedded in that data and project those deficiencies into the future (Hariharan, n.d.).
The first step in tackling health care bias is education, and gender education has to be viewed as a professional discipline rather than just a viewpoint or topic for conversation. Health research and policy integration of sex and gender factors is a multidimensional process. There are unique and developing modalities of involvement within the larger area, despite the fact that there are certain common aims and principles (Yang, 2020). Modern technology may, however, also be used to eliminate disparities that already exist. Bias in health systems may be addressed on an industry-wide level with the use of a sex and gender lens included into data and algorithmic governance. Healthcare data should at minimum contain substantial samples of patients representing different racial and gender identities, particularly in situations where failing to take race or ethnic background into account is linked to known disparities in care.
Data scientists must build curated, public databases with aggregated or de-identified patient data in order to do this. However, this may not always be feasible due to privacy concerns (Thomasian et al., 2021). Still, algorithms capable of handling large data sets could also help support efforts to fill the gap in data by using new research and combining both genetic and epidemiological data to understand gender differences and develop more effective interventions. AI can also be used to reduce care gaps because of the ability to absorb fast-changing medical knowledge, adopt new clinical guidelines and apply or recommend them consistently without the delays and variations inevitable among humans alone. If the full benefits of AI and machine learning are to be realized, it is necessary to have consensus around the regulatory framework of data. Regulation frameworks must incorporate proactive bias testing and AI solutions that promote gender and racial diversity in research, policy and practice in order to address the disparities in healthcare (Hariharan, n.d.; Thomasian et al., 2021).
References:
Cleghorn, E. (2021, June 17). The Long History of Gender Bias in Medicine. TIME. https://time.com/6074224/gender-medicine-history/
Greaves, L., & Ritz, S. A. (2022). Sex, Gender and Health: Mapping the Landscape of Research and Policy. International journal of environmental research and public health, 19(5), 2563. https://doi.org/10.3390/ijerph19052563
Hariharan, K. (n.d.). How Will AI Affect Gender Gaps in Health Care? Marsh McLennan. https://www.marshmclennan.com/insights/publications/2020/apr/how-will-ai-affect-gender-gaps-in-health-care-.html
Thomasian, N. M., Eickhoff, C., & Adashi, E. Y. (2021, November 22). Advancing health equity with artificial intelligence — PMC. NCBI. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8607970/
Wood, S. (n.d.). History of Women’s Participation in Clinical Research | Office of Research on Women’s Health. Office of Research on Women’s Health. https://orwh.od.nih.gov/toolkit/recruitment/history
Yang, H.-C. (2020, September 9). What Should Be Taught and What Is Taught: Integrating Gender into Medical and Health Professions Education for Medical and Nursing Students. NCBI. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7558635/
