Electronic Medical Record System And Data Mining For Assisted Therapy Of Chronic Respiratory Disease

Chronic respiratory disease, as a commonly occurring condition, is often of little concern to people. Common chronic respiratory diseases include: chronic bronchitis, lung trachea dilation, bronchial asthma, pulmonary tuberculosis, lung cancer,etc. Due to the mechanism of formation of chronic respiratory disease not having been completely clarified in pathological research, one of the important clinical demands of any respiratory department is how to clinically identify different chronic respiratory diseases therapy by using clinical tests and prescription data.With the increasing application of the Internet of Things in medicine, and the development of medical engineering demonstrations, to obtain medical knowledge from rich clinical data has been an important aspect of global research on medical informatics. New discoveries in medicine are often derived from analysis of clinical medical data. At present, the data relating to clinical diagnosis and treatment are mainly collected and developed by individuals. The method for medical inheritance of data depends on personal experience, ability, and knowledge so that much valuable medical knowledge cannot be brought to a logical conclusion, or be subjected to continued scientific study. Especially, in the current era of “big data”, it is particularly important to choose a scientific, fast, accurate data analysis method.The author planned to explore the rules of association and clustering of clinical data for chronic respiratory disease to mine more knowledge therefrom. On this basis, new algorithms were proposed through literature analysis which was realised by programming. Moreover, it was verified through simulated experiments and clinical data analysis. Finally, part of the clinical identification of chronic respiratory diseases was realised to meet actual clinical requirements of the respiratory department to some extent.The main work and characteristics of this research are as follows:Firstly, an electronic medical record(EMR) system was designed and implemented. Besides, the author established a method for obtaining the clinical medical data of patients with chronic respiratory diseases for data mining research purposes.The relevant medical data were collected and retrieved by the proposed EMR system and tools, i.e., the studio tool in the Ensemble platform was used to write programs to extract prescription, and clinical laboratory test, data for patients with chronic respiratory diseases. The data sources relating to clinical diagnosis and treatment used in this research were elaborated.Secondly, the author proposed particular Apriori algorithm for the association rules model exploring the diagnosis and treatment information of chronic respiratory diseases. Moreover, the algorithm was used to analyse the data values of chemical index properties in a clinical laboratory for chronic respiratory disease symptoms and the relationships between each main data value thereof to illustrate that the algorithm has practical significance for studying clinical data.Thirdly, the author proposed an improved fuzzy clustering algorithm. In view of the defects in the traditional fuzzy clustering algorithm, such as the amount of calculation required, and its poor clustering effect, the proposed method was improved as follows:(1) Number-field transformation was introduced to the initial data set and the clustering results after each loop iteration. Although introducing the number-field transformation increased the number of calculation steps, it improved the convergence rate of the whole algorithm to raise the computational efficiency in general.(2) Owing to the weighting coefficient m influencing the clustering effect,fuzzy decision-making tools were needed before operating the clustering algorithm to analyse the value of the weighting coefficient m so as to optimise the final clustering results.(3) To reduce the influence of isolated points on the clustering process, a method for selecting the initial clustering centre based on hierarchical clustering was proposed. The selected initial clustering should be as close to the final clustering as possible to decrease the computational effort.(4) Traditional fuzzy clustering algorithms failed to take the number of clusters into account, while in this method, the optimal number of clusters was calculated based on the validity function of the granularity principle. Finally, a standard IRIS data set was used for subsequent computer simulations. By experiments, the improved algorithm and traditional fuzzy clustering algorithms were compared with regard to their accuracy and computational efficiency of clustering. It was proved that the improved algorithm could increase the accuracy of clustering data samples. In addition, the rapid convergence rate could also raise the computational efficiency. Meanwhile, the validation results of clinical data demonstrated that the improved fuzzy clustering algorithm played an obvious role in improving judgment efficiency and rationality when analysing the treatment data of chronic respiratory diseases and when selecting auxiliary diagnosis methods.