We recently spoke with Ma'n H. Zawati, an Associate Professor at 9I制作厂免费 and the Research Director of the Centre of Genomics and Policy. He shared insights on his D2R-funded research, discussing how AI can be used responsibly and ethically to identify genomic biomarkers. He also talked about the potential for personalized therapies and RNA research and the important ethical, legal and policy challenges that need to be addressed to ensure fair healthcare for everyone.
-
Can you explain in simple terms what your project is about?
Our project 鈥�Development of Responsible AI for Genomic Biomarker Identification in the Context of the D2R Initiative鈥� is focused on exploring how artificial intelligence (AI) can be used responsibly and ethically in the context of D2R . We鈥檙e particularly investigating how AI can help identify biomarkers that signal the presence or progression of diseases, while ensuring that this technology is applied in ways that are fair, equitable, and socially responsible.
Biomarkers are biological indicators of disease鈥攅ssentially, signs like genetic changes or patterns that are mutations or variations in gene expression and are linked to a specific disease. By studying these, we can better understand the root causes of illnesses, enabling us to diagnose diseases faster, develop targeted treatments, and even prevent them altogether. While AI technology can now help establish earlier diagnosis and more effective treatments, it comes with important ethical, legal and policy challenges. Will AI models worsen inequities in access to genomic biomarker identification services, particularly for underserved and equity-seeking populations due to increasing costs and technical feasibility? Will machine and deep learning algorithms trained on biased or unrepresentative genomic data further entrench the risks of discrimination that come with the finding of an abnormal biomarker, especially for equity-seeking populations including women, racialized minorities and particularly Indigenous populations, those with disabilities, members of the 2SLGBTQIA+ community, and/or the elderly? These are some of the critical research questions our project seeks to address by mapping out these risks and proposing practical solutions to integrate AI responsibly into genomic and RNA research.
-
How is AI being used in genomic biomarker identification and why is it important for developing personalized therapies for diseases like cancer and rare disorders?
AI is transforming the way we identify genomic biomarkers, particularly by overcoming the limitations of traditional approaches. Historically, biomarker discovery has faced significant challenges, including the use of small sample sizes, difficulties with reproducibility and interpretability. These limitations have hindered progress, especially in understanding and treating complex diseases like cancer and rare disorders. Using AI can be seen as a solution to these challenges thanks to its ability to analyze massive and diverse datasets at an incredible speed. By integrating various types of data, such as genetic sequences, imaging data, clinical information, and more, it can uncover patterns and connections that would be impossible, or at least very difficult, for humans to detect.
This capability is especially important for developing personalized therapies. By identifying specific genomic biomarkers linked to diseases, AI enables researchers to tailor treatments to individual patients鈥� genetic profiles. In the context of cancer, this could mean developing therapies that target unique genetic mutations responsible for tumor growth, in a way that improves effectiveness while minimizing side effects. For rare disorders, AI helps uncover critical biomarkers even when patient populations are small, paving the way for research and treatments that might otherwise remain out of reach.
-
What are the risks associated with using AI in this area?
As mentioned, putting AI to work in genomic biomarker identification holds incredible promise, but it also presents several risks that must be addressed to ensure its responsible and equitable application. One significant concern is bias in data. Many AI models are trained on genomic datasets that predominantly represent individuals of European descent. When these models are applied to underrepresented populations, such as Indigenous, Asian, or African groups, they can sometimes produce inaccurate or even harmful results, which has the further consequence of perpetuating disparities in healthcare outcomes and undermines trust in AI technologies. Privacy risks are another critical issue. Large genomic databases are attractive targets for cyberattacks, as shown by heavily mediatized breaches in direct-to-consumer companies. There are also challenges related to equity in healthcare. AI-driven genomic tools often require significant resources, from advanced computational infrastructure to costly datasets, making them inaccessible to underserved communities. Indigenous and rural populations, for example, may face accessibility barriers to benefiting from these innovations, further widening existing health disparities. Consent challenges further complicate the ethical use of AI in genomics. Participants may not fully understand that their genomic data could be used to train commercial AI systems or shared with third parties. Lastly, reproducibility and interpretability issues pose risks to the reliability of AI-generated findings. Some AI models, especially those using unsupervised learning techniques, can identify genomic patterns without providing clear explanations nor pathways to its results. This "black box" nature makes it difficult for researchers to validate results and for clinicians to confidently apply biomarker discoveries in medical practice.
Addressing these risks is crucial to ensuring that AI technologies are used responsibly and equitably in genomic research and healthcare. The goal is thus to balance innovation with ethical considerations.
-
How does your project aim to address ethico-legal and policy concerns?
By focusing on responsible and equitable integration of AI in this field, we are working to align innovation in genomic research with both societal values and legal standards.
The scope of our work includes several key objectives. First, we will identify the social, ethical, and legal risks associated with using AI for genomic biomarker identification, such as issues of bias, privacy, and equity. To drive actionable change, we will then develop a framework for responsible and fair usage of AI technologies when analyzing genomic and multi-modal health data. Our methods are multifaceted and inclusive. That include policy and legal analysis, engagement with experts through interviews and running online focus groups with participants from equity-seeking populations.
By combining rigorous research, expert input, and public engagement, we seek to provide robust, actionable guidance for the ethical and equitable use of AI to identify genomic biomarkers. Ultimately, our goal is to build trust, safeguard rights, and ensure that these technologies serve the broader public good.
-
You鈥檙e working with a focus group from diverse backgrounds. How will their input help shape the guidelines for using AI in genomic research?
The input from focus groups representing diverse backgrounds will play a pivotal role in shaping our guidelines for the responsible use of AI in genomic research. By engaging directly with equity-seeking populations, like women, racialized minorities (including Indigenous individuals), people with disabilities, members of the 2SLGBTQIA+ community, and the elderly, we are directing our guidance framework to reflect the concerns, values, and lived experiences of those who are often underrepresented in both research and policymaking. This will allow us to build trust and address concerns by these communities, explore key ethical and social dimensions, such as explainability, accountability, and governance. By actively weaving participants鈥� voices into the final ethico-legal guidance document, we are adopting an innovative and inclusive approach. Unlike many existing AI guidelines, which often lack representation from equity-seeking groups, our project prioritizes these perspectives as a cornerstone of the framework.
Ultimately, the diverse insights gathered from these focus groups will ensure that the guidelines we produce are not only legally and ethically sound, but also socially relevant and inclusive. This approach reflects our commitment to creating responsible AI frameworks that address the needs and rights of all communities, particularly those most vulnerable to the risks and inequities of emerging technologies.
-
How do you ensure that the use of AI in healthcare doesn鈥檛 worsen existing inequalities, especially for underrepresented groups?
Ensuring that the use of AI in healthcare doesn鈥檛 worsen existing inequalities鈥攅specially for underrepresented groups鈥攊s a core priority of our project. We are tackling this challenge through a deep commitment to equity at every level of the research process, from the composition of our team to the frameworks guiding our work.
Our team itself embodies diversity and inclusion. This diversity equips us with the lived experiences and nuanced understanding necessary to identify and address issues of bias, inequity, and discrimination in the design and application of AI in healthcare. This diversity isn鈥檛 just a feature of our team: it鈥檚 foundational to our ability to approach this work with sensitivity and insight.
We approach our research recognizing that individuals can experience multi-layered disenfranchisement鈥攂ased on race, gender, ability, socioeconomic status, and more鈥攖hat influences their access to healthcare and the impact of AI-driven tools. Our goal is to help dismantle these inequities by centering the perspectives of underrepresented and underserved communities.
Through our focus groups and engagement with equity-seeking populations, we actively incorporate the perspectives of those who are most affected by health disparities. Their input directly informs the ethico-legal guidelines we鈥檙e developing, ensuring these frameworks promote fairness, transparency, and accessibility for all communities鈥攏ot just those who are already well-represented.
Left to Right: Ma鈥檔 H. Zawati, Associate Professor; Kate Bornais, Research Intern; Yuan Stevens, AI law expert and former-Academic Associate
-
Looking ahead, what do you hope will be the impact of your work? How does your research contribute to D2R鈥檚 vision of delivering genomic-based RNA therapies?
Looking ahead, we hope that our work will have a transformative impact on the responsible and equitable use of AI in genomic research. By identifying ethical, legal, and social best practices for applying AI to genomic biomarker discovery, our project aims to fill critical gaps and set new standards for integrating AI into this rapidly advancing field.
Our research directly aligns with and supports D2R鈥檚 vision of delivering genomic-based RNA therapies. The guidance we develop will provide D2R researchers with the tools and frameworks they need to use AI responsibly in identifying and validating genomic biomarkers. These best practices will inform crucial processes such as biomarker target selection and classification of gene expression patterns, ensuring that the research is both innovative and aligned with ethical and equity principles.
By equipping D2R researchers with a robust set of ethico-legal guidelines, we aim to empower them to lead in the responsible and equitable use of AI for genomic biomarker identification. This leadership will set a precedent for other institutions and initiatives, demonstrating how cutting-edge technology can be harnessed for precision medicine while upholding the highest standards of social responsibility.
Our work will be a vital step in advancing RNA-based therapies, a cornerstone of D2R鈥檚 vision for precision medicine. By ensuring that AI is used effectively and equitably to identify biomarkers, we pave the way for more accurate and personalized RNA therapies. This will lead to treatments that target the root causes of diseases, improving outcomes for diverse populations and addressing health disparities.
In sum, we are very excited by the transformative potential this research holds and look forward to the innovative solutions it could drive in genomic-based RNA research and therapy!
