背景：在诊疗数字化和医学人工智能的背景下，临床决策支持系统（clinical decision support system，CDSS）通过各种人机交互方式在临床的各个环节协助医生诊治具体病人，如准确诊断、选择合适的辅助检查、选择最佳治疗方案等。CDSS有望解决医疗资源紧张、医疗质量不均、医护工作负担重等一系列挑战。然而，CDSS的研发和应用仍有诸多挑战，较为突出的是缺少具有可推理的知识库以支持复杂多样的临床诊疗活动；系统功能及界面设计不友好；鲜有了解医生的真实需求以增加系统的可用性；缺少对系统持续的评估以提高可用性。室性心动过速诊疗复杂疑难，应当研发相应的CDSS进行决策支持。

目的：本研究以可推理的数字化知识建模、符合临床需求的功能和界面设计为核心方法建构CDSS；开发可用于心律失常疾病诊疗的CDSS，并对其进行评估。

内容：本研究内容可分为两大部分，分别是框架研究和案例研究。框架研究包括：①可用于临床决策支持的具有通用性和可推理性的知识引擎；②CDSS功能及界面框架。案例研究以室性心动过速为例，包括：明确室性心动过速诊疗现状及决策支持需求，设计、开发针对室性心动过速的CDSS，并对系统进行评估。形成需求、设计、开发、评估的闭环。

方法：通过文献调研、专家咨询、团队讨论的方法构建知识引擎和决策支持系统的框架。通过专家咨询、问卷调查和访谈的方法研究室性心动过速诊疗现状及心内科医生的决策支持需求。通过设计验证、结构验证和软件开发来研发室性心动过速CDSS。通过模拟仿真法对室性心动过速CDSS进行评估。

成果：形成了一套集成临床信息模型、诊疗模型、陈述性知识和规则性知识，且具有通用性和可推理的知识引擎框架。形成了一套通用的CDSS的功能和界面框架，基于此设计了CDSS原型。了解心内科医生对室性心动过速的诊疗知识和技能掌握程度，并形成一套针对室性心动过速诊疗的决策支持需求。在CDSS原型的基础上，结合室性心动过速的诊疗需求，开发了室性心动过速CDSS，经评估室性心动过速CDSS的诊断正确率超过80%。

结论：本研究形成了通用的可推理知识引擎和CDSS功能及交互界面框架。针对室性心动过速进行需求分析、CDSS设计、开发和评估，从知识和需求两方面提高CDSS的可用性，为专病CDSS的开发和应用提供借鉴。

Background: In the context of healthcare digitization and medical artificial intelligence, clinical decision support systems (CDSS) assist physicians in various clinical aspects through various human-computer interaction methods, such as accurate diagnosis, selection of appropriate auxiliary examinations, and determination of optimal treatment plans. CDSS holds promise in addressing a series of challenges in healthcare, including shortages of medical resources, uneven healthcare quality, and heavy workloads for medical staff. However, the development and application of CDSS still face many challenges, including the lack of inferable knowledge bases to support complex and diverse clinical diagnosis and treatment activities; unfriendly system functionality and interface design; limited understanding of physicians' real needs to enhance system usability; and the absence of continuous system evaluation to improve usability. The diagnosis and treatment of ventricular tachycardia are complex and challenging, necessitating the development of corresponding CDSS for decision support.

Objective: This study aims to construct a CDSS based on inferable digital knowledge modeling, and functional and interface designs that meet clinical needs. The study also aims to develop and evaluate a CDSS for the diagnosis of ventricular tachycardia.

Content: The study can be divided into two main parts: framework research and case study. The framework research includes: 1) the development of a universal and inferable knowledge engine for clinical decision support; 2) the functional and interface framework of the CDSS. The case study focuses on ventricular tachycardia, including: understanding the current status and decision support needs of ventricular tachycardia diagnosis and treatment, designing and developing a CDSS for ventricular tachycardia, and evaluating the system. This forms a closed-loop of requirements, design, development, and evaluation.

Methods: The framework of the knowledge engine and CDSS was constructed through literature review, expert consultation, and team discussion. The current status and decision support needs of ventricular tachycardia diagnosis and treatment among cardiologists were studied through expert consultation, questionnaire surveys, and interviews. The CDSS for ventricular tachycardia was developed through design verification, structural validation, and software development. The usability of the CDSS for ventricular tachycardia was evaluated through simulated simulation methods.

Results: A framework integrating clinical information model, diagnosis and treatment models, declarative knowledge, and procedural knowledge, with a universal and inferable knowledge engine, was developed. A general functional and interface framework for CDSS was established, and a CDSS prototype was designed based on this framework. The study identified the diagnostic and treatment knowledge and skill levels of cardiologists for ventricular tachycardia, and formed a set of decision support requirements for ventricular tachycardia diagnosis and treatment. Based on the general CDSS prototype system, a CDSS specifically for ventricular tachycardia was developed, and in evaluations, the diagnostic accuracy of the CDSS for ventricular tachycardia exceeded 80%.

Conclusion: This study has developed a universal inferable knowledge engine and a CDSS functional and interactive interface framework. It conducted requirement analysis, CDSS design, development, and evaluation specifically for ventricular tachycardia, aiming to improve the usability of CDSS from both knowledge and requirement perspectives, and provide reference for the development and application of disease-specific CDSS.