HuLP: Enhancing Cancer Prognosis with Real-Time Doctor Collaboration
Saudi Arabia is making significant strides in AI-powered healthcare, with two researchers from the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) developing tools that align closely with the Kingdom’s Vision 2030 healthcare goals. Mohammed Firdaus Ridzuan and Tooba Tehreem Sheikh are creating intelligent, interpretable, and scalable AI systems designed to support clinical decision-making in Saudi hospitals.
Ridzuan’s innovation, called Human-in-the-Loop for Prognosis (HuLP), addresses a critical challenge: predicting cancer survival outcomes more accurately by involving doctors directly in the AI process. Unlike conventional AI models that operate independently, HuLP allows medical professionals to adjust predictions based on their clinical expertise in real-time. These adjustments don't just affect individual cases—they help the AI system learn and improve continuously.
This collaborative approach ensures predictions are transparent, context-aware, and tailored to patient-specific data, which is vital for hospitals integrating AI tools at scale. By enabling clinicians to intervene dynamically, HuLP creates a system that adapts to local healthcare realities and builds trust through explainability.
Building Trust and Local Relevance in Saudi Healthcare
Saudi Arabia’s healthcare institutions are undergoing digital transformation through efforts from entities like the Saudi Data and AI Authority (SDAIA) and the Ministry of Health. These organizations focus on integrating AI systems smoothly into clinical workflows. HuLP fits this vision by empowering clinicians to guide AI predictions actively, bridging the gap between AI capabilities and clinical expertise.
One persistent challenge in AI for healthcare is the “black box” problem—where AI decisions are not transparent. HuLP’s design addresses this by making the AI’s reasoning accessible and editable by doctors. Moreover, the system draws on local patient data, which is essential for producing accurate, region-specific predictions that reflect unique health profiles in the Gulf region.
The system's continuous learning from clinician feedback makes it especially suited for real-world deployment, where data consistency can vary. This feedback loop strengthens the AI’s reliability and relevance in clinical practice.
Med-YOLOWorld: A Universal AI Imaging Tool for Multiple Scan Types
While Ridzuan focuses on outcomes, Tooba Tehreem Sheikh is advancing medical imaging with Med-YOLOWorld—an AI system capable of reading nine different types of medical scans at up to 70 frames per second. This system supports modalities including ultrasound, dermoscopy, microscopy, histopathology, CT, and MRI, providing a versatile solution for hospitals that require fast, multi-modal imaging analysis.
Unlike traditional AI tools limited to specific scan types, Med-YOLOWorld uses open-vocabulary detection. This means it can identify anomalies or structures it wasn't explicitly trained on, making it highly scalable and adaptable to new clinical needs.
Overcoming Technical Challenges for Clinical Integration
Sheikh faced significant challenges managing diverse preprocessing needs across scan types. For example, CT and MRI require intensity normalization, while ultrasound and microscopy present different visual characteristics. To address data imbalance—especially for less common imaging types—she developed custom augmentation methods to ensure consistent performance.
Looking forward, Sheikh is integrating Med-YOLOWorld with vision-language models that can explain findings in natural language. This combination offers an end-to-end diagnostic solution that not only detects but also describes and localizes medical issues, enhancing clinical decision support and medical training.
This approach reduces dependency on large annotated datasets, which can be scarce in fast-growing healthcare systems. It also supports detection of emerging diseases or anomalies without needing retraining, a valuable feature for healthcare providers across the Gulf and the broader Middle East and North Africa region.
AI Innovation Supporting Saudi Arabia’s Healthcare Future
Though based in the UAE, MBZUAI alumni like Ridzuan and Sheikh are influencing AI healthcare solutions throughout the Gulf. Their work exemplifies how practical, regionally aware AI tools can support the Kingdom’s push for smart hospitals, real-time imaging, and personalized care.
With ongoing collaboration between research institutions, healthcare providers, and government bodies, systems like HuLP and Med-YOLOWorld could soon be deployed in Saudi hospitals. These tools represent a shift towards AI that is explainable, interactive, and adapted to local clinical needs—helping doctors deliver more accurate diagnoses and predictions.
For healthcare professionals interested in expanding their AI expertise, exploring specialized courses can provide valuable skills to engage with these emerging technologies. Resources like Complete AI Training’s healthcare-focused courses offer practical learning paths tailored to clinical applications.
Your membership also unlocks:
