Construction of prostate cancer diagnosis model based on high throughput sequencing data

VernacularTitle:基于高通量数据集的前列腺癌诊断模型的构建
Author: Funeng JIANG ¹ ; Xin ZHANG ; Chao CHEN ; Zhaodong HAN ; Yongding WU ; Weide ZHONG ; Yuxiang LIANG
Author Information

1. 510180,广州医科大学附属广州市第一人民医院泌尿外科广东省临床分子医学及分子诊断重点实验室
Keywords: Artificial neural network; Genetic algorithm; Diagnosis model; Area under curve
From: Chinese Journal of Urology 2017;38(z1):61-63
CountryChina
Language:Chinese
Abstract: Objective We used the dataset base on high throughput sequencing data to construct a diagnosis model by ANN and GA.Methods We screened the Taylor_prostate datasets from GEO according to,then we used the GA to screen the datas further. Finally we used the ANN to analyze the datas and construct a diagnosis model. To validate the model,we used 10-folds crossvalidation as the inner validation and the datas from Grasso dataset( GPL6480 and GPL6848) were used as the outter validation.Results We got 5 genes ACADL,ACTG2, CACNA2D1,PCP4 and SPARCL1.And we used spss to get the AUC of the model which is 94.62.The result of validation is good.Conclusion The performance of the model is good because the AUC is larger than 0.5.