基于改进离群点检测算法的妇科病案编码数据异常检测研究

Research on Abnormal Detection of Gynecological Medical Record Coding Data Based on Improved Outlier Detection Algorithm

针对如今妇科病案编码数据分析准确度不高的问题,研究提出了融合自组织映射神经网络和局部异常因子的改进检测算法,并将朋友关系模型融入改进算法中得到融合算法。对研究提出的融合算法进行性能对比分析,结果显示,该算法的识别率和误判率分别为97.5%和0.18%,均优于对比算法。随后对该算法进行实证分析后发现,该算法在数据量为3000时的平均时间开销为73秒,显著优于对比算法。上述结果表明研究提出的融合算法具有良好的异常数据检测性能,将其应用于妇科病案编码数据异常检测中可有效提高妇科疾病的诊断准确率。

In response to the problem of low accuracy in analyzing gynecological medical record coding data nowadays,an improved detection algorithm combining self-organizing mapping neural network and local anomaly factors was proposed,and the friend relationship model was integrated into the improved algorithm to obtain the fusion algorithm.The performance comparison analysis of the fusion algorithm proposed in the study shows that the recognition rate and misjudgment rate of the algorithm are 97.5%and 0.18%,respectively,which are superior to the comparison algorithm.After conducting empirical analysis on the algorithm,it was found that the average time cost of the algorithm was 73 seconds when the data volume was 3000,which was significantly better than the comparison algorithm.The above results indicate that the fusion algorithm proposed in the study has good abnormal data detection performance,and its application in abnormal detection of gynecological medical record coding data can effectively improve the diagnostic accuracy of gynecological diseases.