The document discusses how artificial intelligence can both magnify and help address existing biases related to sex and gender in healthcare. It notes that evidence shows sex and gender differences in conditions like diabetes, cardiovascular disease, and cancer. AI tools like big data analytics, natural language processing, and robotics could integrate these differences to improve clinical decisions, but may also perpetuate biases in health data and systems if not designed carefully. The document calls for approaches like explainable AI, bias detection and mitigation, and inclusive design to help ensure AI maximizes well-being for all patients regardless of sex or gender.

1. Sex and Gender
Biases
in Artificial Intelligence
The Impact of
Claudia Falchi Delgado

2. https://www.womensbrainproject.com/
https://www.dexai.eu/
Designing Ethical, Compliant and
Explainable Artificial Intelligence
Our mission is to bring precision to medical
research, towards sustainability in healthcare
The Women's Brain Project DEXAI - Artificial Ethics

3. Guess what? We are
different.

4. Sex AND Gender
differences
Inter-individual differences in health due to
Biological - Genetic (sex) Socio-cultural (gender)

5. diabetes
cardiovascular disorders
neurological diseases
mental health disorders
cancer
autoimmunity
physiological processes such as brain aging
sensitivity to pain
diet
physical activity
tobacco use
alcohol consumption
Moreover, differences in lifestyle factors that are associated with sex and gender.
Evidence of sex and gender differences has
been reported in chronic diseases such as:
(Cirillo et al., 2020)

6. Open questions regarding health differences across the gender
spectrum, due to scarcity of studies dedicated to intersex,
transgender and nonbinary individuals.
Consequences
Observed sex and gender
differences in health and well-
being are influenced by complex
links between both biological and
social-economic factors, which
are often surrounded by
confounding variables such as
stigma, stereotypes, and the
misrepresentation of data.
Consequently, health research
and practices can be entangled
with sex and gender inequalities
and biases.
(Cirillo et al., 2020, p. 2)

7. Artificial Intelligence

8. Artificial intelligence
1 2
As a double-edged sword
AI can magnify and perpetuate existing sex
and gender inequalities
If appropriately designed, mitigate
inequalities by effectively integrating sex and
gender differences in healthcare

9. Artificial Intelligence
Ethics?

11. The difference between them is found in the impact that
these biases have on the patients’ wellbeing and
healthcare access.
Biases
Desirable bias
01 02 Undesirable bias

12. (Cirillo et al., 2020, p. 3)

13. Sources and type of
health data
Experimental and clinical
data
Digital biomarkers

14. Technologies for analysis and
development of health data
Big Data analytics
Natural Language
Processing
Robotics

15. Sex differences in behavioral and social patterns related with
communication such as the number and duration of phone calls
and the degree of social networking callers have been observed.
Big Data analytics
Common Big Data analytics processes and approaches include
the creation of data management infrastructures and the
application of data-driven algorithms and AI solutions.
Associations with mental health and social networks, show
men express higher negativity and lower desire for social
support on social media than women.
Findability, Accessibility,
Interoperability, and Reusability (FAIR)
recommendations for responsible
research and gender equality.
It will facilitate the identification of
sex and gender differences in health,
accurate indicators for prevention and
diagnosis, and effective treatment.
(Cirillo et al., 2020)

16. Allow to make predictions that can contribute to clinical decisions,
such as diagnosis, prognosis, risk of relapse, and symptomatology
fluctuations in response to treatments.
Natural Processing Languages
(NLP)
NLP consists of computational systems that understand and manipulate
written and spoken human language for purposes like machine translation,
speech recognition and conversational interfaces.
Medical chatbots include Woebot, proven to relieve anxiety and
depression, and Moodkit, which recommends chatting and
journaling activities through text and voice notes.
A flourishing area of NLP
is that of medical
chatbots, aiming to
improve users’ wellbeing
through real-time
symptom assessment and
recommendation
interfaces.
(Cirillo et al., 2020)

17. Neurology, Rehabilitation, and assistive approaches for improving
the quality of life of patients and caregivers.
Robotics
Robots are expected to provide personalised
assistance to patients according to their specific needs
and preferences, at the right time and in the right way.
It has been demonstrated that the outcome of a humanoid
robot’s task can be affected by its gender, as in the case of
female charity robots receiving more donations from men
than women.
Awareness of sex and gender
differences in patients and
robots could lead to better
healthcare assistance and
effective human-machine
interactions for biomedical
applications, as well as a better
translation of ethical decision-
making into machines.
(Cirillo et al., 2020)

18. The Digital Divide
Across LMICs (“Overall”), women are 8% less likely than men to own a
mobile phone.
(Cirillo et al., 2020, p. 7)

19. Explainable Artificial
Intelligence (XAI)
In the context of Precision Medicine, the expected outputs of
AI models consist of predictions of risk and diagnosis of
medical conditions or recommendations of treatments, with
profound influence in people’s lives and health.
Where?
XAI is also useful in basic
research, for instance, efforts
in creating “visible” deep
neural networks that provide
automatic explanations of the
impact of a genotypic change
on cellular phenotypic states.
It would enable us to find potential mistaken conclusions
derived by training an algorithm with misrepresented
data -> facilitating the identification of undesirable
biases generally found in clinical data with unbalanced
sex and gender representation.
Explaining the decisional processes will help discover sex
and gender differences in clinical data that is
representative, promoting the -> desired biases for
personalised preventative and therapeutic interventions.

20. A no-code-movement
approach?
Benefits
The development and application of FAIR approaches will be critical
for the implementation of unbiased and interpretable models for
Precision Medicine.
The “no-code-movement” allows non-programmers to
create software using a user-friendly graphical
interface instead of with complex code.
Technology should enable and
facilitate the creation and
understanding and not be a barrier
to entry. In this regard, user
experience (UX) approaches can be
taken to make it inclusive, adaptable,
and customized for each sex and
gender variation representation.
The use of visualizations, logical statements, and
dimensionality reduction techniques can be implemented
in computational tools to achieve interpretability for all
and everyone (patients, researchers, and health
professionals).

21. Recent developments in bias
detection and mitigation
include adopting re-sampling,
adversarial learning, and open-
source toolkits such as IBM AI
Fairness 360 (AIF360).
Reccomendations
01
02
03
05
Distinguish between desirable and undesirable biases and
guarantee the representation of desirable biases in AI
development
Increase awareness of unintended biases in the scientific
community, technology industry, among policy makers, and
the general public
Implement explainable algorithms, which not only provide
understandable explanations for the layperson, but which could also
be equipped with integrated bias detection systems and mitigation
strategies, and validated with appropriate benchmarking
Incorporate key ethical considerations during every stage of
technological development, ensuring that the systems maximize well-
being and health of the population
04
Utilize the "no-code-movement" approach to visualizations, logical statements,
and dimensionality reduction techniques to achieve interoperability for all and
everyone (patients and health professionals)

22. Thank You
Cirillo, D., Catuara-Solarz, S., Morey, C., Guney, E., Subirats, L., Mellino, S., Gigante, A., Valencia, A., Rementeria, M.
J., Chadha, A. S., & Mavridis, N. (2020). Sex and gender differences and biases in artificial intelligence for
biomedicine and healthcare. npj Digital Medicine, 3(1). https://doi.org/10.1038/s41746-020-0288-5