AI for women’s health?

Historically, biomedical research has focused primarily on males: clinical trials have mainly used men as test subjects, and experimental models have mainly relied on male rodents. Results obtained in such limited populations have then been generalized to women, even for illnesses that affect women more often than men. Women’s health advocates have demonstrated that this bias has had detrimental effects, including delayed diagnoses, less effective treatments, and more adverse drug reactions in women.

Efforts that seek to advance women’s health often isolate sex and gender from other inequalities, separate and prioritize sex over gender as a target of investigation, and conceptualize sex (and sometimes even gender) in binary terms. Such reductions of complexity make it easier to raise awareness about women's health, and increase the feasibility of research in this area. However, these tendencies also treat women as a monolith, which harms those who differ from the mean; they marginalize people who are intersex, transgender, and/or nonbinary in particular; and they impede a rigorous understanding of how health disparities emerge in social contexts. Thus, while an approach that moves the needle from ‘one size fits all’ to ‘two sizes fit all’ may be a much needed improvement of the status quo, it also represents a limited progress towards health equity, while also risking the legitimization of stereotypes and inequality beyond the medical sphere.

Within this situation, women’s health advocates hope that artificial intelligence (AI) can advance our understanding of sex/gender health disparities. However, we lack deep insight into how AI techniques may either reify or destabilize human categories when they are used to, for example, discover medical differences between women and men. To address this, this project draws on feminist Science and Technology Studies to ask: Which views on the nature of gender and sex, and with views on the promises and pitfalls of AI, are shaping the discourse and research practices surrounding women's health? And how are categories and facts pertaining to sex, gender, and women then constructed and legitimized through the use of medical AI?

Alongside this research project, a 2-day expert meeting is organized to bring together an international community of scholars who work at the intersection of critical AI studies and feminist/queer theory, in order to discuss the current status of AI and women's health research.