Adoption of Artificial Intelligence in Health Care

Original Title: Navigating the digital health landscape from artificial intelligence-driven molecular subtyping towards optimized rare sarcoma trial design Journal: International journal of surgery (London, England) DOI: 10.1097/JS9.0000000000003040 Overview This correspondence discusses a deep learning framework developed by He and colleagues for the molecular subtyping of leiomyosarcoma using histopathological images. The original study introduced the LMS_DL model, which analyzes single hematoxylin and eosin whole-slide images to predict molecular subtypes. This model achieved an area under the receiver operating characteristic curve (AUROC) of approximately 0.944. Furthermore, the researchers established a prognostic algorithm for predicting two-year overall survival, yielding an AUROC of approximately 0.937. The letter emphasizes how these technical achievements can be…

AI’s Patient Images Show Demographic Biases One-Sentence Summary This study reveals that leading AI text-to-image generators produce patient depictions with significant demographic inaccuracies, over-representing White and normal-weight individuals while failing to reflect real-world disease epidemiology. Overview As artificial intelligence (AI) text-to-image generators become widely used for creating visual content, their application in medical contexts raises concerns about accuracy and bias. This research systematically evaluated four popular AI models—Adobe Firefly, Bing Image Generator, Meta Imagine, and Midjourney—to assess how accurately they depict patients for 29 different diseases. Researchers generated a total of 9060 images and had twelve independent raters assess the depicted sex, age, race/ethnicity, and weight. These AI-generated demographics were…

Unveiling System Dynamics with Neural Symbolic Regression One-Sentence Summary The paper introduces a computational tool, Learning Law of Changes (LLC), that combines neural networks and symbolic regression to automatically discover the mathematical equations governing complex network dynamics from observational data. Overview Understanding the behavior of complex systems, such as biological networks or epidemic spreads, is a fundamental challenge in science. These systems are often governed by underlying mathematical rules, typically in the form of differential equations. However, identifying these exact equations from data alone is notoriously difficult. This paper presents a novel computational framework named LLC designed to tackle this problem. The method first employs neural networks to learn the…

Original Title: Benefit-Risk Reporting for FDA-Cleared Artificial Intelligence-Enabled Medical Devices Journal: JAMA health forum DOI: 10.1001/jamahealthforum.2025.3351 FDA AI/ML Device Reporting Lacks Transparency Overview A comprehensive analysis of 691 artificial intelligence and machine learning (AI/ML) medical devices cleared by the US Food and Drug Administration (FDA) between 1995 and 2023 reveals significant deficiencies in benefit-risk reporting. The cross-sectional study examined FDA decision summaries and postmarket surveillance databases. It found that crucial information was frequently missing. For instance, 95.5% of device summaries lacked demographic data for the populations on which the AI was tested, 53.3% did not report the training sample size, and 46.7% omitted the study design. The evidence supporting clearance…

Original Title: Biomarkers Journal: Alzheimer's & dementia : the journal of the Alzheimer's Association DOI: 10.1002/alz70856_106355 Overview Mild cognitive impairment (MCI) serves as a critical transitional stage between the typical cognitive changes of aging and the onset of Alzheimer's disease. This study explores structural brain alterations associated with this condition by quantifying gray matter and white matter volumes using high-resolution T1-weighted magnetic resonance imaging. The research team utilized a specialized deep neural network named Vb-Net to perform automated segmentation and volumetric analysis on healthy controls and individuals with MCI. Patients with MCI experienced a 4.60% reduction in gray matter volume and a 5.60% decrease in white matter volume compared to…

Original Title: Multimodal intelligent prediction model for in vitro fertilization Journal: NPJ digital medicine DOI: 10.1038/s41746-025-02331-5 Overview This study introduces VaTEP, a multimodal deep learning framework that integrates time-lapse system videos of developing embryos with tabular clinical data. Developed and validated using data from 9,786 participants across three medical centers, VaTEP predicts three clinical outcomes: fetal heartbeat presence, singleton versus multiple pregnancy, and miscarriage versus live birth. Using a multi-task learning approach, the system optimizes these predictions simultaneously. Results show the model achieved an area under the curve (AUC) of 0.8000 for fetal heartbeat, 0.8823 for singleton versus multiple pregnancy, and 0.9258 for live birth versus miscarriage. These values exceeded…