Biomedical data, both in ‘traditional’, analogue forms as well as in the form of digital, ‘big’ data, are contingent social products. They reflect the categories and practices that structure our societies. We illustrate this by discussing gender biases in data stemming from clinical trials and electronic health records (EHR) and consider how biomedical data are prone to bias in different phases of data work, from data capture and representation to category building and analysis to using outputs. We argue that developments such as ‘Personalised’ and ‘Precision Medicine’ that have been made possible by ‘big data’ analyses could be seen as a shift away from the male ‘standard patient’ by trying to comprehensively and objectively represent many different aspects of patients’ lives and bodies. At the same time, the very promises of comprehensiveness and objectivity are problematic: The data generated and collected, as well as the infrastructures and analytic tools used to do this, reflect the social realities – including the injustices and inequities – within which they were developed. The knowledge created on the basis of this ‘evidence’ can thus perpetuate existing biases. While we do not subscribe to a view of the world that considers truly objective, neutral, and – in this sense –‘unbiased’ knowledge possible or even desirable, we suggest a number of ways in which gender bias in biomedical data should be made visible, reflected upon, and in certain instances acted upon.
ASJC Scopus Sachgebiete
- Menschliche Einflussgrößen und Ergonomie
- Geisteswissenschaftliche Fächer (sonstige)
- Soziologie und Politikwissenschaften
- Human-computer interaction