In a groundbreaking move towards combating oral cancer, the Faculty of Dentistry at the University of Hong Kong (HKU), in collaboration with the Prince Philip Dental Hospital (PPDH), has inaugurated the world's first artificial intelligence (AI) clinic. Launched in April 2025, this innovative facility is specifically designed to focus on the early detection and prevention of oral cancer by catering to individuals diagnosed with oral potentially malignant disorders (OPMD). These disorders encompass a range of lesions within the oral cavity that carry a risk of developing into oral cancer.
Harnessing AI Power: OralCancerPredict
At the heart of this pioneering clinic lies OralCancerPredict, a bespoke AI-powered web tool meticulously developed by HKU's Faculty of Dentistry in 2021. This tool signifies a substantial leap forward in employing decision-support technology within oral healthcare. Its primary function is to predict the likelihood of cancer development in patients presenting with specific OPMD subtypes, notably oral leukoplakia (OL) and oral lichenoid diseases (OLD). These two conditions are among the most frequently encountered forms of OPMD and are recognized for their elevated potential to transform into oral cancer when compared to healthy oral tissues.
Addressing a Critical Global Health Challenge
Oral cancer stands as the most prevalent malignancy within the head and neck region, presenting a significant global health challenge. Alarmingly, the worldwide five-year survival rate hovers around a concerning 50%. The situation in Hong Kong mirrors this urgency, with approximately 60% of oral cancer cases being diagnosed at an advanced stage. This stark reality underscores the critical need for proactive strategies focused on early detection. The newly established AI clinic, equipped with OralCancerPredict, directly addresses this challenge by offering sophisticated, precision-based risk assessment tools. These tools empower clinicians to identify individuals at higher risk of developing oral cancer at an earlier, more treatable stage.
Leveraging Decades of Data for Enhanced Accuracy
The AI tool's predictive capabilities are rooted in a comprehensive dataset meticulously compiled from the extensive patient records of Queen Mary Hospital, spanning nearly two decades (2003-2022). OralCancerPredict operates by analyzing a multitude of patient-specific data points, including demographics, clinical presentation, detailed histological findings, and any co-existing medical conditions. This multidimensional analysis culminates in the generation of a precise cancer risk probability score. The robustness and generalizability of this model were further validated through its application to patient cohorts from the United Kingdom and Nigeria, affirming its potential for widespread use across diverse populations.
Precision in Treatment Planning and Resource Optimization
The AI-driven technology effectively categorizes patients into high-risk and low-risk groups. For individuals identified as high-risk, the recommended course of action typically involves surgical intervention. Conversely, patients classified as low-risk may be managed through careful observation, patient education and counselling, and supportive therapies. This risk-stratified approach not only streamlines the development of tailored treatment plans but also contributes to the efficient allocation of healthcare resources by potentially avoiding unnecessary surgical procedures for patients with a lower probability of disease progression.
Superior Predictive Accuracy and Clinical Confidence
The algorithm underpinning OralCancerPredict has demonstrated an impressive average accuracy rate of 94%, coupled with high calibration performance. This level of predictive accuracy surpasses that of conventional diagnostic methods currently employed in the assessment of OL and OLD. Consequently, clinicians are provided with a more reliable and robust framework to guide their treatment planning decisions. Furthermore, the model offers an interpretable rationale behind each prediction, enhancing transparency and fostering greater clinician confidence in AI-assisted decision-making processes.
Accessible Innovation for Equitable Healthcare
A significant advantage of this pioneering clinic is its commitment to accessibility. Despite the integration of cutting-edge AI technology, the OralCancerPredict model does not impose any additional costs on patients or necessitate supplementary diagnostic tests. As the AI models are made freely available, treatment decisions remain cost-effective and equitable. Patients can access the specialized services of the AI clinic through a referral from their general dental or medical practitioner.
A Multidisciplinary Team Driving Innovation
The AI clinic is under the leadership of Professor Richard Su, Chief and Clinical Professor of the Division of Oral and Maxillofacial Surgery (OMFS). He is supported by a highly skilled multidisciplinary team comprising Clinical Associate Professor Liwu Zheng, Clinical Assistant Professor Jane Jingya Pu, Dr Joanna Ko from OMFS, and Assistant Professor John Adeoye from the Digital/Precision Dentistry unit. This collaborative synergy, bringing together expertise in surgery, clinical practice, and computational science, ensures the delivery of comprehensive and holistic patient care.
Setting a Global Standard for AI Integration in Healthcare
The incorporation of AI into routine clinical workflows represents a transformative evolution in the delivery of oral healthcare. By embedding sophisticated data-driven tools within a dedicated service model, HKU's AI clinic enhances diagnostic precision, empowers informed clinical decision-making, and facilitates timely and targeted interventions. As healthcare systems worldwide seek scalable and effective solutions for the early detection of cancer, this innovative AI clinic at the University of Hong Kong sets a compelling precedent for the seamless integration of intelligent technologies into mainstream patient care.