Medical information modelling, processing and retrieval for computerized knowledge assistance towards evidence-based clinical practice

Healthcare is complex domain and distributed in nature. Some of its major characteristics are shared and distributed decision making and management of care. Such characteristics require the communication of complex and diverse forms of information between clinical and other settings. The healthcare services centre around the wellness, maintenance, and illness management of patients. The delivery of such services heavily depends on up-to-date medical knowledge and experience. However, miscommunication, misinterpretation of diverse forms of information exchanged between clinical settings, inability to be up-to-date with clinical best practices, and accessing rapidly changing and exponentially growing medical knowledge cause errors in health care that are leading cause of injuries and deaths.In this thesis, we addressed the research problems related to the interpretation and understanding of medical referral and response letters, exchanged between Specialists and General Practitioners (GPs) for patients care decision making. These research problems are implicitly associated with GPs information needs at point of care, including management of and access to evidence-based medical information and knowledge relevant to medical referral and response letters. The interpretation and understanding of referral and response letters and sending alert s for critical situations at point of care require methods for medical information processing, and methodologies for modelling and management of clinical knowledge. It also requires mechanisms to access concise, relevant, evidence-based clinical knowledge and techniques for efficient and context-sensitive retrieval. The goal of the thesis has been divided into five objectives, therefore, five-phased multi-steps research methodology was devised. We have taken interdisciplinary solution approach along the lines of healthcare knowledge management, contextual information retrieval, and knowledge-based search strategies. We have formulated a knowledge modelling methodology and a computerization technique for clinical practice guidelines, which transform them into computer interpretable segments that are enriched with content-specific meta-information. To link segments of computerized clinical practice guidelines with online evidence-based medical knowledge, we have developed a technique for automatically generating clinical queries from these segments. These queries are used by our Context Specific Query Generation framework to retrieve relevant medical literature from online evidence-based knowledge sources and to link them with corresponding computerized CPGs segments. We have developed a method for computerized processing of referral and response letters. This method analyses medical information and provides a comprehensive information-view of the letter 10 help healthcare practitioners formulate customized and focused information specifications to access required knowledge. We also have developed a technique for contextual and statistical analysis of medical concepts and indexing strategy, which are used to retrieve CPGs segments and related medical literature relevant to information needs of healthcare practitioners. Finally, we have designed a healthcare-knowledge mediated architecture and implemented a computer system for Clinical Knowledge Assistance (CKA) to provide better interpretation and understanding of referral and response letters