Nature Article: “Sex and gender differences and biases in Artificial Intelligence for Healthcare”
By Simona Mellino, WBP Vice President
The Women’s Brain Project is proud to announce that an article co-authored by a number of their members, “Sex and gender differences and biases in Artificial Intelligence for Biomedicine and Healthcare” just came out in Nature Digital Medicine.
Below, we present an overview of the work, why it matters, and the details of an upcoming webinar on June 10 where some of the co-authors will present their findings before opening up to Q&A.
Why does it matter?
The objective of Precision Medicine is to tailor the approach to treatment and prevention by considering individual variability and moving away from the “one-size-fits-all” approach. To achieve this, one must account for biologic, lifestyle, and environmental factors.
Both sex (biologic factor) and gender (social factor) are important to account for as they can affect risk factors, prevalence, treatment response, etc. as demonstrated across several diseases.
However, sex and gender differences in health and well-being are influenced by a complex mix of biological and social-economic factors and can be entangled with inequalities and bias. The awareness of these biases has increased over the past years, as demonstrated by a growing body of publications and in part driven by new artificial intelligence (AI) technologies applied across several domains in healthcare.
Whilst these advances offer the opportunity to mitigate biases and inequalities, they might also eventually increase these inequalities if they are not developed removing biases.
Teaser video – Author interview
What did we do for the paper?
We decided to perform a systematic review by looking at two main aspects: sources of health data, and the role of several different technologies. We explored potential biases and contribution to create personalized interventions. After an initial introduction and explanation of different biases, our review covers the following aspects:
Sources of health data
- Experimental and clinical data
- Digital biomarkers
Technologies for the analysis and deployment of health data
- Big Data analytics
- Natural language processing
- Explainable AI (XAI)
Output of health technologies
We conclude with a set of recommendations to ensure that sex and gender differences are accounted for in AI implementations that informs Precision Medicine.
If you’re curious to find out more about the paper from a “backstage” perspective, you can read the Nature Digital Medicine “Behind the Paper” article published on June 11, 2020. Click here to access it.
What do we plan next?
Our recent Webinar (see details below) featured four of the article co-authors: Davide Cirillo, Silvina Catuara Solarz, myself (Simona Mellino), and Antonella Santuccione Chadha. Other co-authors dialed in and provided insights as part of the live Q&A.
The format of the Webinar was:
- Introduction about sex and gender differences and types of biases
- Specific examples related to sources of health data and technologies
- Summary of our recommendation for fair AI development
- Live Q&A session
- Webinar: Sex and gender differences and biases in Artificial Intelligence for Biomedicine and Healthcare
- Time & Date: Wednesday 10 June, 8.30-9.30pm CET
To view the recording of the webinar, click the video below
This review marks the start of our work to further explore sex and gender differences as a pathway towards Precision Medicine. We aim to elaborate on the review at this point in time, have a few specific outputs in mind, including a Hackathon, whose results will be presented at our International Forum on Women’s Brain and Mental Health (19-20 September 2020)
To make sure you don’t miss the webinar or future announcements pertaining to the Hackathon, the WBP Forum, or other news, follow us on social media (we’re on LinkedIn, Twitter, Instagram, and Facebook) or sign up for our mailing list.