How Umaima Rahman is Making Medical AI Safer and Fairer for Hospitals Everywhere
Umaima Rahman develops AI models that perform reliably across hospitals, improving medical imaging accuracy. Her work boosts patient safety and supports clinicians worldwide.
How Umaima Rahman is Making Medical AI Safer and More Reliable
Umaima Rahman, a 29-year-old Indian researcher, recently earned her PhD in Computer Vision from the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) in Abu Dhabi. She is among the first female students to receive a doctoral degree from the university.
Addressing the Challenge of AI Transferability in Healthcare
A key issue in medical AI is that models trained in one hospital often fail when applied in another. Differences in imaging equipment and protocols can cause AI tools to misinterpret medical images, leading to incorrect diagnoses. For example, an AI system trained on high-quality X-rays from a modern hospital might not perform well on images from clinics with older machines.
Rahman explains, “A slight change in imaging protocol can cause an AI model, which otherwise detects cancer with near-perfect accuracy, to fail when used in a different setting.” This problem limits AI’s usefulness in real-world healthcare environments.
Developing AI Models That Generalize Across Settings
To tackle this, Rahman spent four years creating AI models that generalize well across hospitals, scanners, and patient populations. Her approach teaches AI to focus on medically relevant features while ignoring variations in image quality or device type.
She emphasizes the stakes: “In medical imaging, the consequences of poor AI performance can be life-threatening. Reliable AI systems are essential to patient safety and ensuring equitable healthcare.”
Innovating for Rare and Emerging Diseases
Rahman introduced the concept of “cross-disease transferability,” where an AI model trained to detect one disease can assist in identifying others affecting the same organ. This approach is particularly valuable during health crises like COVID-19 and for resource-limited settings where advanced imaging tools aren’t always accessible.
Her research has been shared internationally, including presentations in Switzerland, connecting with medical professionals and AI experts.
Looking Ahead: Continuing the Mission
Currently exploring postdoctoral roles at institutions such as Stanford and MIT, Rahman plans to return to the UAE to advance healthcare AI further. She stresses that AI should support, not replace, clinicians by enhancing their decision-making and improving care quality.
Rahman also hopes to encourage more women to enter AI fields. “There are many women in biomedical engineering but fewer in AI. I want to help change that and eventually become a professor to mentor others.”
For healthcare professionals interested in AI’s evolving role in medicine, exploring specialized AI training can be valuable. Resources like Complete AI Training’s healthcare-focused courses offer practical skills to work alongside cutting-edge technologies.
