Dental plaque is the primary etiological factor for common oral diseases such as dental caries and gingivitis, particularly affecting susceptible populations including children. Preventing plaque-induced dental diseases in pediatric patients remains a global challenge as certain plaque regions are often missed during daily cleaning. Early detection and removal of plaque are essential but often difficult to achieve consistently. A routine monitoring tool that visualizes plaque accumulation on primary teeth could allow parents to actively track their children’s oral health, enabling timely prevention and intervention for dental caries and periodontal diseases. Inspired by these, we present PlaqueSAM, an automatic plaque detection system using intraoral photographs captured by smartphones. By uploading six images captured with the assistance of alignment overlays, PlaqueSAM can generate a comprehensive plaque detection report with visual references. PlaqueSAM was developed using 3402 prospectively collected intraoral photographs from 567 preschool children across 11 kindergartens in Hong Kong using a three-stage training strategy based on a multi-task learning paradigm. The performance of PlaqueSAM was quantitatively evaluated on 81 children, achieving sensitivity, specificity, and accuracy of 76.58, 82.47, and 79.60%, respectively. These findings suggest that PlaqueSAM may support future plaque screening workflows without requiring disclosing agents or plastic retractors typically used by dental professionals, although home-based use requires further validation under real-world user-operated conditions. With further real-world validation, PlaqueSAM may support dynamic monitoring of children’s daily plaque accumulation and cleaning, thereby contributing to the advancement of teledentistry and promoting children’s oral health remotely, conveniently, and at a lower cost.
GenAI and agentic systems in medicine bring real opportunities for better diagnosis and therapy and for the development of better diagnoses and therapies. It is now increasingly being recognised that the power of these systems for good is inseparable from their power to be used for harm. Systems optimised for biomedical research are unfortunately being harnessed by malign human masters to help develop bioweapons to be targeted at humanity. Agentic systems are autonomously, unknown to their human masters, carrying out cybersecurity attacks. Frontier models are increasing anxiety and stress of medical students and threaten to reshape medical professional roles, not always for the better. Although there may currently be some over emphasis of risk by BigTech foundation model developers, and these may even be attempts at ‘regulatory capture’, there are also genuine reasons for concern. We set out actionable measures that are needed to control medical frontier models to maximise human benefits and control risks.
Approximately one in ten live births is preterm ( < 37 weeks gestational age), placing these infants at elevated risk because of immature physiology. Analysis of movement patterns in the Neonatal Intensive Care Unit (NICU) is essential for detecting complications, yet current monitoring approaches remain limited. Using routine photoplethysmography (PPG), we extracted movement bouts from 65 preterm infants (47 without health complications, 18 with severe neurological conditions) across 50 days using a validated algorithm. Movement actigraphy was summarized using three novel metrics: (1) percent time moving per day across postmenstrual age, (2) daily distributions of movement durations quantified by power law exponents, and (3) ultradian movement cycles across 24 hours. While percent time moving did not change with age, power law exponents scaled linearly with age, indicating longer movements with maturation. Daily movement activity revealed 3-hour ultradian cycles corresponding to NICU feeding schedules, alongside an anticipatory rise in spontaneous motor activity before feeding. Infants with neurological complications, including two individual case studies, showed deviations in activity patterns when compared to uncomplicated infants. These results suggest that longitudinal movement tracking using routine sensors reveals systematic activity patterns and offers a scalable clinical tool for evaluation of health in this high-risk population.
Artificial intelligence (AI) is entering every stage of medical education, yet guidance for responsible integration remains limited. We synthesize evidence across undergraduate, graduate, and continuing medical education within institutional, learning, and readiness and safety systems. Integration should reflect task structure, verifiability, consequence, and oversight. Continuing medical education remains comparatively underdeveloped despite its importance for practicing physicians, who use clinical AI directly and shape trainee practices, institutional adoption, and responsible implementation.
