The Application of Data Mining Technology for the Judgment of Poisoning Cases

Through the collection of lots of relating literatures and analysis of over 6000 actual poisoning cases, this paper established a poisoning features model (living) and autopsy findings model (died) to describe the regularities and differences on the poisoning symptoms caused by different poisons. Based on the two models and combined with the application of the data mining technology the poison can be judged only by the poisoning symptoms of the victims before the clinical diagnosis in hospitals and chemical analysis in labs. After the Apriori, SVM and GRI algorithms were compared with the data mining tool (PASW Modeler 13.0, SPSS Inc.), several operational models were built by Apriori algorithm, thus successfully realizing the function of judging the poisons according to the poisoning symptoms only. In recent two years, in the actual application of this Lab and other labs, the accuracy of a single result obtained by the judgment method established is above 50%, and that of the result list of 10 poisons is over 75%.