Abstract:

Doctor recommendation can help patients find telemedicine specialists who meet their needs quickly and accurately at a professional level, and ensure the value of healthcare delivery. An adaptive recommendation method for telemedicine specialists with feedback adjustment was proposed from the perspective of privacy protection, which emphasizes the adaptability and acceptability of recommended results. Through the modeling of patient electronic medical records and specialist long- and short-term knowledge features, similarities between knowledge features were mined, thereby alleviating the cold-start problem in specialist recommendation and generating specialist recommendation indices. Then, the interest and activity of specialists and feedback mechanism were incorporated into the recommendation framework, which realized the dynamic closed-loop adjustment and optimization of specialist recommendation while considering the individual behavior changes of specialists and preferences of patients, making the recommended results interpretable and adaptive, and improving the recommendation ability. Finally, the effectiveness of the fusion recommendation strategy proposed in this paper was verified by conducting multi-group comparative analysis experiments, which achieved the effective adaptation of patients’ needs and service capabilities. The research results not only have guiding significance for the practical telemedicine, but also provide a theoretical reference for future research.

Key words: specialist recommendation, feedback adjustment, interest, activity, long- and short-term knowledge

摘要：

医生推荐能够从专业层面帮助患者快速准确地找到符合需求的医生，确保医疗服务价值的有效实现。从隐私保护出发，提出了一种带有反馈调节的远程医疗专家自适化推荐方法，更强调推荐结果的适应性及可接受性。通过对患者病历和专家长、短期知识特征建模，挖掘知识特征间的相似性，据此缓解专家推荐中的冷启动问题并生成专家推荐指数；将专家活跃度、兴趣度及反馈机制纳入推荐框架，在考虑专家个体行为变化与患者偏好的同时实现专家推荐的动态闭环调整优化，使推荐结果更具解释性和自适性，提升推荐能力；多组对比分析实验验证了融合推荐策略的有效性，实现了患者需求和服务能力的有效适配。研究成果不仅能为远程医疗实践提供借鉴，也为后续相关研究提供理论参考。

关键词: 专家推荐, 反馈调节, 兴趣度, 活跃度, 长短期知识