Can AI Combat Gender Bias in Healthcare? Experts Believe It’s Possible

Artificial Intelligence (AI) has the potential to revolutionise healthcare by assisting doctors in diagnosing conditions and suggesting advanced treatments. However, concerns remain that if AI systems are built on biased data, they may inadvertently reinforce existing gender inequities in healthcare. Experts now argue that AI can also be a tool to fight these disparities—if developed with a gender-sensitive approach.

The Problem: Gender Bias in AI Models

Many AI systems have historically been developed based on data sets that prioritise men’s health needs, reflecting a male-dominated tech industry. This bias poses a risk to women and non-binary patients, who may receive less accurate diagnoses or inappropriate treatments. A study published in PLOS One in 2020 highlighted how Natural Language Processing (NLP) models in psychiatry exhibited gender biases, often under diagnosing women with PTSD while incorrectly diagnosing them with personality disorders.

The issue is not limited to language; it extends to the design of AI tools and medical devices. For example, traditional models for assessing heart attack risk have often overlooked how symptoms manifest differently in women, leading to underdiagnosis and delayed treatment. Additionally, during the COVID-19 pandemic, personal protective equipment (PPE) designed primarily for male bodies exposed how biases in design can have safety implications. Around 70% of global healthcare workers are women, yet PPE was often oversized and ill-fitting for female staff, leading to inadequate protection and increased risk of accidents.

Moving Towards Inclusive Design

The AI industry is gradually acknowledging these issues, with initiatives like Bioinfo4women-B4W and the Australian Centre for Sex and Gender Equity in Health and Medicine leading efforts to address gender bias. Experts believe that inclusive AI development can lead to safer, more effective healthcare solutions by incorporating data diversity and ensuring that algorithms reflect the needs of all genders.

For instance, in 2022, the Global Registry of Acute Coronary Events (GRACE) was updated to include AI predictive models that analyse male and female data separately. This change has led to more accurate heart attack assessments for women, who are now more likely to receive early interventions, demonstrating how AI can address structural biases in patient care.

Practical Solutions: SMARThealth Pregnancy GPT

A recent example of inclusive AI development is the SMARThealth Pregnancy GPT, created by The George Institute for Global Health. This tool was designed to provide guideline-based pregnancy advice to women in rural and remote areas of India, where access to healthcare is often limited. The team collaborated with local community health workers, clinicians, and women from these communities to refine the algorithm, ensuring that the AI chatbot was contextually accurate, culturally sensitive, and free from harmful stereotypes.

The result was an AI system capable of supporting healthcare workers and educating women in under-resourced settings, demonstrating the potential of gender-sensitive AI to improve health outcomes while avoiding entrenched biases.

The Future of AI in Healthcare: Gender-Sensitive Development

As AI becomes an integral part of healthcare, ensuring the quality and integrity of the data used to train these systems is crucial. Gender-sensitive AI could improve medical technologies in numerous ways, including:

- Tailoring personalised treatments by considering sex-specific factors

- Predicting treatment responses more accurately for women and non-binary individuals

- Enhancing diagnostic tools through better data representation

- Supporting virtual healthcare services to provide equitable care remotely

The recently launched UK Medical Science Sex and Gender Equity program advocates for the routine consideration of sex and gender in research, aiming to set a scientific standard that supports the development of fair, effective AI applications in healthcare.

The Ethical Imperative

"AI is the future of healthcare, and we can't afford to repeat the health inequities of the past by ignoring sex and gender," says Dr. Sue Haupt, Senior Research Fellow at the Centre for Sex and Gender Equity in Health and Medicine. "It’s time to program AI to lead us towards an ethical and equitable healthcare system."

Experts are hopeful that, with appropriate focus on inclusivity, AI can not only address but help eliminate long-standing gender biases in healthcare. Tools like SMARThealth Pregnancy GPT offer a glimpse of what is possible when the design and development of AI incorporate diverse, representative data sets. If AI models are developed to account for biological differences and cultural contexts, they can bridge gaps in care, rather than widen them.

The information provided in this article is for educational purposes and should not be taken as medical advice.

Note: Published originally by https://www.downtoearth.org.in/