School of Computer and Information Engineering, Xiamen University of Technology, Xiamen, 361024, China.
Engineering Research Center of Big Data Application in Private Health Medicine, Fujian Provincial University, Putian, 351100, China.
BMC Med Inform Decis Mak. 2022 Aug 13;22(1):218. doi: 10.1186/s12911-022-01963-x.
The clinical practice of shared decision-making (SDM) has grown in importance. However, most studies on SDM practice concentrated on providing auxiliary knowledge from the third-party standpoint without consideration for the value preferences of doctors and patients. The essences of these methods are complete and manual negotiation, and the problems of high cost, time consumption, delayed response, and decision fatigue are serious.
In response to the above limitations, this article proposes a fuzzy constraint-directed agent-based negotiation and recommendation framework for bilateral and multi-issue preference negotiation in SDM (PN-SDM). Its purpose is to provide preference information and intellectualize PN-SDM to promote SDM practice. We modeled PN-SDM problems as distributed fuzzy constraint satisfaction problems and designed the doctor agent and patient agent to negotiate on behalf of the doctor and patient. The negotiation result was then transformed into treatment plans by the recommendation model. The proposed negotiation and recommendation models were introduced in detail by an instance.
The proposed method with different strategies and negotiation pairs achieves good performance in terms of negotiation running time, negotiation rounds, and combined aggregated satisfaction value. Specifically, it can feasibly and effectively complete multiple rounds of PN-SDM in a few seconds and obtain higher satisfaction.
The experimental results indicate that the negotiation model can effectively simulate preference negotiation and relieve the pressure of increasing issues. The recommendation model can assist in decision-making and help to realize SDM. In addition, it can flexibly cope with various negotiation scenarios by using different negotiation strategies (e.g., collaborative, win-win, and competitive).
共享决策（SDM）的临床实践变得越来越重要。然而，大多数关于 SDM 实践的研究都集中于从第三方的角度提供辅助知识，而没有考虑医生和患者的价值偏好。这些方法的本质是完全手动协商，存在成本高、耗时、响应延迟和决策疲劳等问题。
针对上述局限性，本文提出了一种模糊约束导向的基于代理的双边和多议题偏好协商框架用于 SDM 中的偏好协商（PN-SDM）。其目的是提供偏好信息并实现 PN-SDM 的智能化，以促进 SDM 的实践。我们将 PN-SDM 问题建模为分布式模糊约束满足问题，并设计了医生代理和患者代理代表医生和患者进行协商。然后，推荐模型将协商结果转换为治疗方案。通过实例详细介绍了提出的协商和推荐模型。
不同策略和协商对的提出方法在协商运行时间、协商轮数和综合聚合满意度值方面都取得了良好的效果。具体来说，它可以在几秒钟内以合理且有效的方式完成多轮 PN-SDM，并获得更高的满意度。
实验结果表明，协商模型可以有效地模拟偏好协商并减轻议题增加的压力。推荐模型可以协助决策并有助于实现 SDM。此外，通过使用不同的协商策略（如协作、双赢和竞争），它可以灵活应对各种协商场景。
Zotero 插件