Research And Application On The Evaluation Model Of Healthcare Quality Based On Big Data

With the advent of the big data age, data from all walks of life has increased rapidly. As one of the most valuable big data in all industries, medical big data has been studied frequently in recent years. Methods for medical quality evaluation are mostly based on the analysis of the traditional experience and statistical methods. However, with the gradual increase of medical data, traditional medical quality evaluation model shows a lot of problems, such as the sample is too large, the processing is slow and so on. Based on Medicare big data in a region, the study researched and realized a medical quality evaluation model, which integrated the traditional evaluation methods and data mining ideas. The main contents of this study were as follows:1. The establishment of outlier indicator. Medicare data was accurate and effective after noise processing, and the outliers could reflect certain medical quality information. In this thesis, the specific application of KNN algorithm was also carried out, such as pruning and so on, so the time complexity of the algorithm could be reduced and would be more suitable for big data processing. Based on statistical method and improved KNN algorithm, established our outlier indicator, defined as the ratio of each of the two detection algorithm results to the proportion of outliers in each hospital. Experimental results showed that it was a good test of hospitals with poor-quality medical care.2. The establishment of excellent cases rate indicator. We mainly applied the clustering analysis in data mining in the Inclusive Model, changed the traditional empirical dichotomy into the automatic clustering method from the data itself, and established the excellent cases rate indicator which was more accurate. The study used X-means algorithm to automatic clustering, and for the reason that X-means algorithm was not very good in more than four dimensional data, the thesis proposed a new classification method based on the attribute overlap ratio(AOR). And we used clustering for the classification results. Experimental results showed that the accuracy of the Inclusive Model and clustering purity has been greatly improved.3. Medical quality evaluation model. Based on the outlier indicator and excellent cases rate indicator, and calculated by the model formula, the final evaluation score of each hospital was gained. And then we carried on the medical quality evaluation and classification of each hospital. Experimental results showed that this model has a good evaluation effect for medical quality.4. The development of big data medical quality evaluation system. This system was based on Hadoop distributed platform, used sqoop to import data from the Oracle Database, through HDFS and HIVE to data storage and manage, applied outlier indicator and excellent cases rate indicator, and displayed the corresponding results. The results showed that it had a highly practical value.