Evaluation And Validation Of Gastric Cancer Based On Decision Tree Analysis Evaluation And Validation

Objective: This study was based on the difference of data statistics algorithm and traditional syndrome differentiation method.The research is based on the establishment of China has been in the node cockin project database of 1 gastric cancer cases on the basis of formulating relevant evaluation methods,and then use the advanced expert syndrome differentiation of traditional decision tree data statistical analysis method to obtain the patients with gastric cancer type information,the establishment of data model,and use the R language to construct syndrome model validation,finally evaluate the rationality and potency of this analysis method.In order to study the gastric cancer diagnosis standard,but also can provide a reasonable basis for other types of disease differentiation research.Methods: According to the established database of 1 gastric cancer cases of syndrome differentiation of traditional way,leading experts and then uses decision tree method to obtain the operation in patients with gastric cancer classification data,set up for evaluation model,finally to verify the diagnosis model to evaluate its rationality and potency by decision tree method.Finally the expert conclusion of syndrome differentiation with the algorithm of artificial differentiation results were compared,and through the discussion of the results.Conclusion:1.The conclusion based on TCM expert’s traditional differentiation :(1)The syndrome differentiation of whole period(according to the frequency ordering): syndrome of spleen-stomach deficiency-cold,syndrome of deficiency of both qi and blood,syndrome of blood stasis and toxin resistance,syndrome of incoordination between liver and stomach,syndrome of stomach heat injuring yin,syndrome of Phlegm stagnation.(2)Syndrome differentiation in patients before operation(according to the frequency ordering): syndrome of spleen-stomach deficiency-cold,syndrome of incoordination between liver and stomach,syndrome of blood stasis and toxin resistance,syndrome of deficiency of both qi and blood,syndrome of stomach heat injuring yin,syndrome of Phlegm stagnation.(3)Syndrome differentiation in patients after operation(according to the frequency ordering):syndrome of spleen-stomach deficiency-cold,syndrome of blood stasis and toxin resistance,syndrome of incoordination between liver and stomach,syndrome of deficiency of both qi and blood,syndrome of stomach heat injuring yin,syndrome of Phlegm stagnation.(4)Syndrome differentiation in patients after operation and chemotherapy(according to the frequency ordering):syndrome of deficiency of both qi and blood,syndrome of spleen-stomach deficiency-cold,syndrome of blood stasis and toxin resistance,syndrome of stomach heat injuringyin,syndrome of Phlegm stagnation,syndrome of incoordination between liver and stomach.2.Decision tree analysis based on the conclusions:(1)Accuracy of 6 combination of syndrome differentiation according to the importance of modeling variables and expert analysis of patients with gastric cancer were compared with full time decision tree:Ganweibuhe(85.76%),Piweixuhan(83.67%),Yuduneizu(71.66%),Weireshangyin(60%),Qixueliangxu(49.68%),Tanshi zuwei(31.18%).The total accuracy of 67.73%.Time by 50 percent off cross validation showed the accuracy of all time is 67.13%;(2)The preoperative decision tree analysis of 6 combinations of syndrome types according to the importance of modeling variables and expert identification For comparison of accuracy were:Ganweibuhe(89.62%),Piweixuhan(86.54%),Weireshangyin(52.94%),Yuduneizu(43.75%),Qixueliangxu(43.33%),Tanshizuwei(0%).The total accuracy of 67.57%.time by 50 percent off cross validation showed gastric cancer the accuracy rate was 63.65%;(3)The decision tree analysis of postoperative gastric cancer the accuracy of 6 combinations of syndrome differentiation according to the importance of modeling variables and expert comparison respectively:Weireshangyin(88.89%),Ganweibuhe(88.24%),Piweixuhan(76.67%),Yuduneizu(76.32%),Qixueliangxu(71.74%),Tanshizuwei(55%).The total accuracy of 77.96%.Time by 50 percent off cross validation showed the accuracy rate of postoperative gastric cancer was 69.74% times;(4)The decision tree analysis of chemotherapy after operation of gastric cancer after accuracy of 6 combination of syndrome differentiation according to the importance of modeling variables and expert comparison respectively:Piweixuhan(92.39%),Yuduneizu(81.25%),Ganweibuhe(80.15%),Weireshangyin(58.97%),Tanshizuwei(45.94%),Qixueliangxu(43.59%).The total accuracy of this period of time was 68.92%.After 50 percent off cross validation showed that the accuracy rate was 70.62%.3.The decision tree analysis method in this paper according to the analysis of database research has dialectical classification skills good data found in the analysis of the conclusions in model analysis is demonstrated,and the good fitting degree,is a feasible and effective TCM standards of the tool.