Artificial Neural Network Based Method for Classification of Gene Expression Data of Human Diseases along with Privacy Preserving

Abstract

In this paper, the author introduces a classification approach using  Artificial Neural Network(ANN) with Back-Propagation learning technique for human diseases like Cancer and heart problems from clinical diagnosis data. Clinical diagnosis is done mostly by experienced doctors with expertise in this field. In many cases, the test results are not effective towards the diagnosis of the disease. The author is particular about the wrong diagnosis which leads to a wrong treatment. The author is using Artificial Neural Network technique to classify the disease with reduced number of DNA sequence. The accuracy is differing based on the training data set and validation data set.  The other major issue is the privacy preserving of the patients. As we are sharing the critical data from clinical diagnostic centers, there is good chance of patient’s anonymity is revealed. To avoid this, the author is using a simple Privacy Preserving in Data Mining (PPDM) technique to crypt the identity of the patients as well as the critical data and discloses only the required data like DNA sequence to the research team, as they are not much interested in the identity or the owner of the diagnosis report.