AI发现特定心脏病准确率已超过人类专家？心电图迎来技术飞跃

Core Insights - Early detection of structural heart disease is crucial for improving prognosis, but widespread screening is limited by the cost and accessibility of imaging tools like echocardiography [1][2] - The recent study published in Nature demonstrates that an AI tool named EchoNext can independently analyze ECG data to accurately identify patients with structural heart disease, outperforming human experts [1][3] Group 1: AI Tool Development - EchoNext was developed to analyze routine ECG data, effectively screening for patients who may require further echocardiography, thus optimizing medical resource allocation [1][2] - The AI model was trained on over 1.2 million pairs of ECG and echocardiography data from 230,000 patients, enabling it to detect various forms of structural heart disease [2] Group 2: Performance Metrics - EchoNext identified over 7,500 high-risk patients from a cohort of 85,000 who had not previously undergone echocardiography, achieving a diagnostic accuracy of 77.3% for structural heart disease [2][3] - In a comparative study, human cardiologists had an accuracy of 64% without AI assistance, which improved to 69.2% with AI support, but still fell short of EchoNext's performance [3] Group 3: Clinical Implications - Structural heart disease affects millions globally and includes conditions like valvular heart disease and heart failure, with early detection linked to reduced mortality and healthcare costs [4] - The integration of ECG and AI could revolutionize screening practices, allowing clinicians to better determine which patients should undergo echocardiography [4] Group 4: Future Directions - Future heart disease risk prediction may benefit from multimodal models that incorporate data from chest X-rays, lab tests, and ECGs for comprehensive risk assessment [4] - Challenges remain in the integration and adoption of AI models in clinical settings, necessitating further optimization and refinement [5]