MBZUAI's Vision for AI-Driven Healthcare Transformation in the UAE
Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), the world's first graduate-level research university dedicated exclusively to advancing artificial intelligence (AI), is at the forefront of integrating AI into medicine. Located in Abu Dhabi, United Arab Emirates (UAE), MBZUAI leverages its expertise to address pressing healthcare challenges through innovative AI solutions. These efforts focus on early disease detection across all life stages—from fetal development to elderly care—aligning with the UAE's national agenda for precision medicine and digital health transformation.
The university's BioMedIA lab and Institute of Digital Public Health play pivotal roles, collaborating with local institutions like Cleveland Clinic Abu Dhabi and the Department of Health (DoH) Abu Dhabi. This synergy not only accelerates research but also ensures practical deployment in UAE hospitals, tackling prevalent issues such as diabetes affecting 16% of adults and rising dementia cases projected to burden global healthcare by 2050.
Revolutionizing Fetal Health Monitoring with Ultrasound AI
One of MBZUAI's groundbreaking contributions begins even before birth. Associate Professor Mohammad Yaqub's team developed ScanNav, evolving into FetalCLIP and MobileFetalCLIP models trained on over 210,000 fetal ultrasound images. These AI systems automate anomaly detection, such as heart defects, which affect one in 33 global births annually according to the World Health Organization (WHO).
The process works step-by-step: First, the model processes ultrasound frames to segment fetal anatomy. Next, it measures key biometrics like head circumference and heart structures. Finally, it flags abnormalities with high precision, enabling earlier interventions. MobileFetalCLIP runs on edge devices, making it accessible in remote UAE clinics via partnerships with GE Healthcare. This reduces diagnostic time from hours to seconds, crucial in a country where congenital anomalies demand rapid response.
In the UAE context, where expatriate populations bring diverse genetics, these tools standardize screening, minimizing human error and supporting the nation's maternal health goals.
Non-Invasive Screening for Chronic Diseases Using Retinal Imaging
Transitioning to adulthood, MBZUAI's oculomics research uses simple eye scans to detect systemic diseases early. Assistant Professor Jianing Qiu's work analyzes retinal images for signs of diabetes, hypertension, Alzheimer's, and cardiovascular risks. In demonstrations at Cleveland Clinic Abu Dhabi, these AI models combined retinal data with electrocardiograms (ECGs) to predict heart failure.
- Retinal vessel changes indicate hypertension with 90%+ accuracy.
- Microaneurysms signal diabetic retinopathy before symptoms.
- Optical coherence tomography (OCT) reveals Alzheimer's biomarkers up to a decade early.
This multimodal approach processes images through convolutional neural networks (CNNs), extracting features like vessel density and nerve fiber layers, then fuses with ECG waveforms via transformers for holistic predictions. In the UAE, where diabetes prevalence is high, such tools enable population-scale screening at primary care levels, preventing complications and reducing healthcare costs by up to 30% through timely management.
AI Tackling Cancer: From Detection to Prognosis
MBZUAI alumnus Numan Saeed, now a research scientist, leads efforts in oncology AI. The HECKTOR project uses multi-modal datasets from 10+ global centers, including UAE hospitals like Sheikh Shakhbout Medical City, for head and neck cancer segmentation from CT/PET scans. Achieving superior tumor delineation, it supports staging and prognosis.
A larger initiative, funded by DoH, builds a Cancer Foundation Model for breast, colorectal, and head/neck cancers using electronic health records (EHRs), imaging, and genomics. Step-by-step: Data preprocessing aligns modalities; foundation models pre-train on vast datasets; fine-tuning yields personalized risk scores. Collaborations with Burjeel and Corniche Hospitals ensure UAE-specific data, addressing local epidemiology like rising colorectal cases.
Early detection via these models could boost UAE cancer survival rates, mirroring global trends where AI halves diagnostic delays.
🧠 Predicting Neurodegenerative Diseases Decades in Advance
For aging populations, Ph.D. student Salma Hassan's MICCAI 2025 presentations introduced ClinGRAD and MAGNET-AD. ClinGRAD, a multimodal graph neural network, classifies dementia subtypes—Alzheimer's, vascular dementia, mild cognitive impairment—with 98.75% accuracy on the ANMerge dataset, outperforming baselines by fusing MRI, genomics, and clinical data. It offers interpretability via node importance, aiding clinicians.
MAGNET-AD, a spatiotemporal GNN, predicts Preclinical Alzheimer's Cognitive Composite (PACC) scores and conversion time with a 0.8582 concordance index, using longitudinal MRI/fMRI, genetics, and EHRs. By modeling temporal progression, it forecasts onset up to 20 years early, vital as dementia cases may reach 152 million globally by 2050. Codes are open-sourced on GitHub for reproducibility.ClinGRAD GitHub MAGNET-AD GitHub
In UAE's growing elderly demographic, these tools promise proactive care, reducing misdiagnosis rates up to 30%.
Overcoming Language Barriers with Multilingual Medical AI
Dr. Hisham Cholakkar's BiMediX models and Arabic Doctor app bridge accessibility gaps. Supporting Arabic, English, and Hindi, it provides preliminary diagnostics via text/voice, with 140,000+ Hugging Face downloads. EMNLP/MICCAI publications validate its clinical reasoning, empowering underserved Middle East/Africa regions.
This aligns with UAE's diverse population, enhancing equity in healthcare delivery.
Strategic Partnerships Accelerating AI Deployment
MBZUAI's MoU with DoH Abu Dhabi targets genomics and AI diagnostics for prevention/early detection. Events with Cleveland Clinic showcased robotic ultrasound, agentic AI for monitoring, and multi-omics analytics by Prof. Nataša Pržulj.
These collaborations embed AI in UAE's ecosystem, from Sheikh Shakhbout to Burjeel, fostering data-sharing and ethical AI governance.
Drug Discovery and Simulation: The AIDO Platform
GenBio AI's AIDO, UAE AI Award 2025 winner, simulates biology via DNA/RNA/protein models like GET transformer. Predicting molecule effects accelerates drug discovery, reducing costs and timelines for UAE-specific needs like diabetes therapeutics.
Challenges, Ethical Considerations, and Future Outlook
While promising, challenges include data privacy (addressed via federated learning), bias mitigation in diverse UAE populations, and clinician integration. MBZUAI emphasizes explainable AI (XAI) for trust.
Future: Human Phenotype Project maps UAE genetics; School of Digital Public Health scales public strategies. By 2030, expect AI routine in UAE clinics, positioning MBZUAI as global leader and boosting higher education's role in national innovation.
- Expand datasets with UAE biobanks.
- Deploy MobileFetalCLIP nationwide.
- Integrate MAGNET-AD in elderly screening.
This positions UAE universities like MBZUAI as hubs for AI-health fusion, attracting talent and driving economic growth.
