Toronto-based JMIR Publications has just released a new batch of “News and Perspectives” pieces that spotlight how fast-moving health technology research is colliding with everyday clinical reality—and with AI at the center of medical education. The coverage spans three themes: AI-assisted learning, foundation models for rare disease biology, and a noninvasive “electronic nose” approach to cancer detection through volatile compounds in skin-exhaled breath.
In “Charting a Course for AI in Medical Education,” Katie Cottingham examines how medical schools are adopting AI tutors and feedback systems. The practical promise is clear: tools can provide rapid, personalized coaching and help students practice decision-making. But the article stresses a fundamental educational risk—students may become dependent on AI answers rather than developing the cognitive struggle that makes knowledge durable. The piece argues for “scaffolding” rather than replacement, using AI to guide reasoning while preserving the integrity of clinical thinking.
The rare disease report, “AI Models Could Improve Diagnosis and Care for Rare Diseases,” focuses on data scarcity and underdiagnosis. With rare disorders affecting hundreds of millions worldwide, machine learning faces a harsh constraint: limited labeled datasets. Simon Spichak describes deep generative approaches such as popEVE, a foundation-style model designed to learn patterns in genetic variation and propose candidate pathogenic variants.
Crucially, the article notes that these models can surface novel genetic signals—even when the biomedical literature is thin. Yet it also cautions that impressive offline results have not always translated into robust real-world diagnostic performance. Bridging that gap likely requires careful validation, diverse cohorts, and alignment with clinical workflows.
Finally, Liam Critchley reports on progress in e-noses for cancer screening in “Advanced Olfactory Cancer Detection: When E-Noses Sniff the Skin.” Instead of invasive sampling, the approach targets volatile organic compounds carried through skin-exhaled breath.
The pilot study highlighted quantum-dot-based sensing using cadmium sulfide nanocrystals to capture chemical signatures. Researchers report high discrimination between cancer patients and healthy controls, with the potential to classify disease severity—though the clinical validation pathway remains essential before deployment.
Together, these pieces paint a viral, forward-looking picture of digital health: AI can accelerate discovery and learning, but only if systems are engineered for trust, calibration, and meaningful human oversight.
Subject of Research: People; medical diagnosis; medical genetics; artificial intelligence; education/education technology; clinical training; biomarkers.
Article Title: Charting a Course for AI in Medical Education; AI Models Could Improve Diagnosis and Care for Rare Diseases; Advanced Olfactory Cancer Detection: When E-Noses Sniff the Skin.
News Publication Date: July 17, 2026.
Web References: https://www.jmir.org/2026/1/e106582
References: Cottingham K. Charting a Course for AI in Medical Education. J Med Internet Res 2026;28:e106486. DOI: 10.2196/106486. Spichak S. AI Models Could Improve Diagnosis and Care for Rare Diseases. J Med Internet Res 2026;28:e106852. DOI: 10.2196/106582. Critchley L. Advanced Olfactory Cancer Detection: When E-Noses Sniff the Skin. J Med Internet Res 2026;28:e106481. DOI: 10.2196/100948.
Image Credits: N/A.
Keywords: artificial intelligence; medical education; foundation models; rare diseases; electronic nose; cancer screening; biomarkers; clinical training; volatile organic compounds.
