Research On Semantic Clinical Decision Support Model Integrating Patient Preferences

With the continuous development of information technology in China,information technology has been basically integrated into all aspects of our lives.Clinical Decision Support System(CDSS)aims to assist medical decision-making through targeted clinical knowledge,patient information and other health characteristics,so as to improve the decision-making suggestions provided by medical care.The clinical decision support system based on evidence-based medicine matches the characteristics of each patient with the computerized clinical knowledge base,and then provides the evaluation or suggestions suitable for patients to clinicians to assist them in clinical decision-making.However,the related clinical decision support system still faces several thorny problems.CDSS is difficult to integrate into the existing workflow of clinical experts.Many CDSS exist independently as bulky systems and cannot communicate with other systems effectively.The interruption of work flow will lead to the increase of doctors’ original workload,and then increase the time of diagnosis and shorten the time of face-to-face consultation with patients.As the brain of clinical decision support system,knowledge base needs to keep pace with medical practice literature and clinical guidelines.The latest information needs to be integrated into the system to maintain the value of the system.Clinical decision-making requires doctors to implement evidence-based diagnosis and treatment activities in a complex data environment,which may lead to diagnosis and treatment bias caused by lack of knowledge.In addition,evidence-based medicine draws evidence from randomized controlled trials based on coarse-grained groups.Therefore,it is impossible to make an accurate and objective evaluation of the individuals who do not belong to the scope of the original experiment.Even if the clinical decision-making evidence of patients with special cases is included,patients may have special preferences different from others.In order to solve the above problems,the research proposes a semantic clinical decision support model integrating patient preferences.For the research of semantic clinical decision support model,aiming at the problem of insufficient interaction between clinical decision support system and data layer,a clinical decision knowledge base model(PCKB)which supports semantic interoperability and integrates patient preferences is proposed.Ontology and semantic rules are used to realize the clear,abstract and standardized expression of clinical decision-making domain knowledge.In the aspect of integrating patients’ preferences,we take the experience data of similar patients from social networks as subjective evidence,crawl patients’ comments from social networks and conduct emotional analysis to form Patients’ Preference Ontology(PPO).On the basis of reusing the existing ontology,combined with the patient preference ontology,the class and its hierarchy are defined.Owl semantic expression is used to realize semantic modeling of knowledge base to ensure the continuous updating and improvement of knowledge base.Through the ontology based knowledge base,the semantic mapping between EHR and clinical decision support model is established on the basis of following the international standard of information model in medical and health field.The structured patient data is transformed into a shared semantic data representation.Through semantic mapping,the information path between clinical decision support system and EHR is opened up,which provides real-time and reliable data support for diagnosis and treatment behavior.Ontology reasoning mechanism is used to infer the knowledge base,and SWRL clinical decision rules are written to infer the decision path,so as to generate the final recommendation decision.A Rule Inference Engine(RIE)is designed and developed to implement decision rules and reasoning logic,match patient data with clinical knowledge in the knowledge base,and generate objective diagnosis and treatment suggestions.Then,VAS is used to elicit patients’ preferences,and RIE is used to match similar experience cases with patients’ preferences,and the best recommendation decision is given.An interactive decision-making interface is designed and developed to collect patient information and present recommended decision-making results.The interactive knowledge graph is used to express decision-making evidence more intuitively to help clinical experts understand the reasons behind decision-making and enhance their cognition of clinical arguments.A CDSS prototype has been developed to implement this method,and the feasibility of our framework is illustrated by a case study of triple assessment of breast cancer.