Breast Cancer Prediction by Logistic Regression with CUDA Parallel Programming Support
Received Date: Dec 14, 2015 / Accepted Date: Mar 22, 2016 / Published Date: Jul 30, 2016
Objective: The present article shows the development and the simulation of a machine learning model created with logistic regression to predict breast cancer tumor.
Methods: The software is developed under Linux Ubuntu, with Theano Framework. It uses Python programming language and Nvidia CUDA parallel GPU programming mechanism. It uses Nvidia CUDA programming approach to take advantage of multiple GPUs.
Results: From the results we can say that the model is very efficient. We developed two versions. The first version gives, in 85% of cases, the right response while last and more optimized version is able to give 91% of good responses. They are significant values and the differences between the versions may open better scenario for the future.
Conclusion: The good responses of the method developed could be open better scenario for breast cancer disease to avoid, sometimes, invasive diagnostic analysis. Furthermore, with a different sample of study is possibile to improve the efficiency of the methods mixing some different input dataset. Create a web database to train the algorithm behind the model to create a sort of open data for consultations.
Keywords: Breast cancer; Bioinformatics; Logistic regression model; CUDA parallel programming; supervised algorithm; Machine learning; Data analysis
Citation: Peretti A, Amenta F (2016) Breast Cancer Prediction by Logistic Regression with CUDA Parallel Programming Support. Breast Can Curr Res 1: 111. Doi: 10.4172/2572-4118.1000111
Copyright: ©2016 Peretti A, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Select your language of interest to view the total content in your interested language
Share This Article
- Total views: 9733
- [From(publication date): 9-2016 - May 22, 2018]
- Breakdown by view type
- HTML page views: 9620
- PDF downloads: 113