20230427-华安证券-医疗AI专题二_专业性与实用性兼备_ChatGPT类聊天AI在医疗端的使用报告_18页_1mb

ChatGPT exhibits strong professionalism in medical applications, with GPT-4 achieving USMLE accuracy of 78.63%, outperforming GPT-3.5, and demonstrating capability in handling complex, multi-disciplinary medical cases across various regions and specialties. It shows potential as a reliable tool for medical consultation, but requires enhancements for Chinese-specific contexts, such as improved accuracy for local medical queries and integration with image/data analysis.

In practicality tests, AI chat models like ChatGPT-4, GPT-3.5, and competitors (e.g., NewBing, Wenxin Yiyan) provide accurate, user-friendly responses in medical inquiries, such as hypertension management, comparable to doctor advice. For instance, ChatGPT-4 offers comprehensive lifestyle and monitoring suggestions, enhancing patient experience, while other models may be more cautious or rely on external searches.

Applications include significantly increasing healthcare accessibility and quality in consumer-facing and primary care settings, enabling better remote medical advice and support. This could upgrade internet healthcare services and improve services in underserved areas, potentially boosting efficiency and affordability in the long term.

Investment opportunities lie in internet healthcare companies (e.g., JD Health, AliHealth) and primary healthcare providers (e.g., Rundata Medical, Yifeng Pharmacy) that leverage AI for improved services. Risks involve regulatory changes, AI development uncertainties, and inherent limitations requiring human oversight.