Global Genomics Summit at MBZUAI Explores AI’s Transformative Role in Precision Medicine
International researchers and healthcare leaders convene at Mohamed bin Zayed University of Artificial Intelligence to address the critical underrepresentation of non-European populations in genetic research and advance equitable, AI-driven precision medicine globally.
Mohamed bin Zayed University of Artificial Intelligence recently hosted the Global Genomics Summit, an occasion of considerable significance that brought together distinguished international researchers and healthcare leaders to examine the profound implications of artificial intelligence and expansive genetic datasets for advancing precision medicine, an increasingly sophisticated approach that tailors medical care to the unique characteristics of individual patients.
The summit’s deliberations centred upon three fundamental questions: how to ensure the inclusion of non-European populations whose genetic information remains substantially underrepresented in global research endeavours; how artificial intelligence might enhance the interpretation of genetic information; and how scientific discoveries can be translated into equitable clinical applications that serve humanity’s diverse populations with fairness and efficacy.
The current landscape of genetic research reveals a concerning imbalance. As of September 2026, approximately 88% of participants in genome-wide association studies, the comprehensive investigations that establish connections between genetic variations and disease, derive from individuals of European ancestry. Arab populations, who constitute roughly 5 to 6% of the world’s population, represent less than 1% of participants in these foundational studies. Research demonstrates compelling benefits when diverse populations are incorporated into the development of polygenic risk scores, with such diversity enhancing predictive accuracy across all population groups.
Carlos Bustamante, summit co-chair and Acting Director of Research Translation, emphasised that artificial intelligence models require training data reflecting genuine global population diversity. He observed that even substantial sequencing efforts confined to a single nation would inevitably leave significant gaps in understanding. “One of the big challenges in trying to build good AI models is access to sufficient data to train and the most important thing we want to guard against is bias in the training data,” Bustamante noted, whilst calling upon the global research community to collaborate in addressing these critical deficiencies and establishing the relationships essential for meaningful scientific partnerships.
The summit’s comprehensive programme encompassed sessions exploring genomic infrastructure in East Asia, genomics networks throughout Africa, and biobanking, the systematic preservation of biological samples and health information for research purposes. Participants also engaged in substantive discussions regarding disease prediction through analysis of medical records and genetic information, polygenic risk scores that estimate individual susceptibility to disease, and artificial intelligence models designed for interpreting rare genetic variations.
Yulia Medvedeva, summit co-chair and Assistant Professor of Computational Biology at MBZUAI, observed that the expanding availability of genetic and biological data creates unprecedented opportunities for developing artificial intelligence models with practical applications in research and precision medicine. She emphasised the value of sharing knowledge regarding biobank and data bank operations, whilst stressing the paramount importance of trust and properly authorised access to sensitive genetic information. “Trust is important when working with sensitive genetic data and conducting research based on it,” Medvedeva explained, pointing to federated learning as a particularly promising avenue for advancing collaboration. This innovative technique permits artificial intelligence models to learn from multiple biobanks whilst preserving the confidentiality of underlying sensitive data, provided datasets maintain compatible structures, alongside opportunities for exchanging research tools to enhance the utility of artificial intelligence applications.
Source: WAM