الفهرس | Only 14 pages are availabe for public view |
Abstract Electronic Medical Records (EMR) can be defined as an organized collection of electronic health information about patients and populations. Electronic format facilitates information sharing across different health care sites, and easies its embedding in network-connected enterprise-wide information systems like medical Case Based Reasoning systems (CBR). The sustained and ubiquitous availability of high-quality Operable Clinical Cases (OCC) is deemed as a bottleneck towards the incorporation of medical CBR systems in any real life medical diagnostic environment. Procurement of CBR compliant cases is quite challenging, as this requires medical experts to map their experiential knowledge to an unfamiliar computational formalism. Although there are many EMR sources (files – databases) over the internet and in patient-care places, the presence of well-structured EMR files in the internet is a major problem for researchers in medical informatics, and integrating different EMR databases is also another challenge. To overcome the above two problems in order to get EMR, two techniques are proposed: The first one is an efficient technique to extract EMR from different databases, it is based on retrieving different relationships between patients’ different data tables (files) and automatically generates EMRs in XML format, then building frame based medical cases to form a case repository that can be used in medical diagnostic systems. This technique has been applied on different structured databases, taken from DAR EL FOAAD hospital sited in the 6th of October city and EL TAYSEER hospital sited in EL Sharkia city. Generated EMR has been evaluated by medical experts based in these two hospitals and they have confirmed that the generated EMR are accurate with percentage 85 %. The second technique is an intelligent approach for knowledge transformation from EMR to clinical cases. This technique is based on Case-Based Reasoning methodology. CBR can be used in developing medical diagnostic systems. It targets to collect medical documents from Internet (MEDLINE library), and convert collected EMR document to clinical cases. It consists of two main phases: First phase is based on using Semantic Similarity Retrieval Model (SSRM) in retrieving medical documents. SSRM overcomes the semantics problem, and associates retrieved documents containing semantically similar terms. It has been tested on OHSUMED (This test collection was created to assist information retrieval research. It is a clinically-oriented MEDLINE subset, consisting of 348,566 references. The test collection was built as part of a study assessing the use of MEDLINE by physicians in a clinical setting). All OHSUMED documents are indexed by title, and symptom. Second phase is using retrieved medical documents to generate medical cases by mapping these EMR document data to defined cases attributes. Finally building a case base repository that would be utilized in a CBR-medical diagnostic system. The experiment results (30 documents) had been done using Vector Space Model (VSM) and SSRM with expansion with very similar terms (T = 0.9, T = 0.7, and T = 0.5). For larger answer sets, SSRM with expansion threshold T = 0.5 found to be the best method. For short answer sets, SSRM with expansion threshold T = 0.5 is the best method. An explanation may be that it introduced many new terms and not all of them are conceptually similar with the original query terms. Generated cases have been manual validated by domain experts based on (El Tayseer Hospitals) in the following domains (blood tests, general Surgery) and reported that 80 % of generated cases are accurate. They also reported that the rest percentage 20 % of generated cases have missed a lot of data and there are repetitive values that make these cases incorrect. |