Core Insights - The article discusses the latest advancements in AI applications within the medical and health sectors, highlighting its potential to assist in disease diagnosis and treatment, including innovative methods such as using earwax to detect Parkinson's disease [3]. Group 1: AI in Weight Management - The Shanghai Jiao Tong University team published a study on June 23, 2025, in Nature Medicine, introducing the REVERIE system, an AI-based virtual reality sports system aimed at helping overweight adolescents lose weight through a randomized controlled trial [3][4]. - REVERIE combines deep reinforcement learning and Transformer architecture, featuring a virtual coach that provides personalized guidance, achieving metabolic effects comparable to real exercise while enhancing cognitive and motivational aspects [4][7]. - The study demonstrated that VR exercise is as effective as real exercise in fat reduction and metabolic improvement, while also significantly enhancing psychological health and cognitive abilities [7]. Group 2: AI in Precision Medicine - On July 14, 2025, a collaborative study by Imperial College London and Spanish scientists published in Nature Communications introduced AIDA, an AI-driven gastric inflammation diagnostic assistant that achieved a 94.1% eradication rate for Helicobacter pylori, surpassing traditional methods [8][11]. - AIDA integrates clinical, molecular, and imaging data to create a patient-centered decision-making system, facilitating personalized follow-up and treatment recommendations [11]. - The project involves collaboration among 15 European centers, ensuring compliance with scientific, ethical, and legal standards, thereby enhancing the tool's reliability [11]. Group 3: AI in Endoscopic Diagnosis - A study published on June 20, 2025, by Fudan University in Lancet Digital Health focused on an AI model for detecting nasopharyngeal carcinoma in endoscopic images, significantly improving diagnostic accuracy for ENT specialists [12][14]. - The AI model increased diagnostic accuracy from 83.4% to 91.2% and reduced the time taken to analyze images, demonstrating high clinical applicability [14][15]. - The model was trained on a large dataset from 42 hospitals across 24 provinces, showcasing its robustness and potential for widespread application in similar diagnostic scenarios [15]. Group 4: AI in Prognostic Modeling - On July 2, 2025, a team from University College London published a study in Lancet Digital Health validating a multimodal AI-derived prognostic model for advanced prostate cancer patients, utilizing data from four phase 3 clinical trials [16][18]. - The AI model, MMAI, demonstrated a significant correlation with prostate cancer-specific mortality rates, providing detailed stratification of patient risk levels [18]. - MMAI enhances the accuracy of prognostic assessments for advanced prostate cancer, offering a practical tool for patient management and treatment decision-making without requiring additional tests [18].