Scientists train deep-learning models to scrutinize biopsies like a human pathologist
Peer-Reviewed Publication
Updates every hour. Last Updated: 15-Dec-2025 10:11 ET (15-Dec-2025 15:11 GMT/UTC)
In order to improve the diagnostic accuracy of deep-learning AI algorithms, models require larger amounts of high-quality training data, which presents a significant burden for pathologists or radiologists that diagnose disease based on images composed of billions of pixels. Researchers have developed a method to imitate the expertise of pathologists by tracking their eye movements while diagnosing whole slide images. This data helps scientists train AI models to more accurately identify regions of interest and better classify tissue samples based on the behavior of highly experienced and trained professionals with little to no additional burden placed on these providers.
Researchers have developed a new, data-driven way of fitting prosthetic legs which could lead to better fitting prosthetics, in less time and at a lower cost.
Researchers demonstrated a broadband infrared frequency comb that can operate stably, efficiently, and accurately without the need for bulky external components. The device could be utilized in a remote sensor or portable mass spectrometer that can track and monitor multiple chemicals in real-time for extended periods.
An artificial intelligence (AI)-assisted model that combines a patient’s MRI, biochemical, and clinical information shows preliminary promise in improving predictions of whether their knee osteoarthritis may soon worsen. Ting Wang of Chongqing Medical University, China, and colleagues present this model August 21st in the open-access journal PLOS Medicine.