Personalized Liver Cancer Risk Prediction Using Big Data Analytics Techniques with Image Processing Segmentation.

Area of Science:

• Computational biology
• Bioinformatics
• Data science

Background:

• Drug discovery and disease prediction face challenges with time-consuming, computationally intensive methods.
• Virtual screening (VS) utilizes big data techniques but requires significant processing for ligand docking and protein receptor analysis.
• Accurate and efficient prediction of diseases like liver cancer is crucial for developing new treatments.

Purpose of the Study:

• To develop an efficient big data analytics approach for liver cancer prediction.
• To introduce a novel, scalable algorithm for processing large datasets in disease analysis.
• To improve the performance of machine learning methods in drug discovery and medical diagnostics.

Main Methods:

• The study employed image processing for cancer segmentation and feature extraction.
• Big data analytics techniques, including MapReduce and Mahout algorithms, were used for prefiltering ligand data.
• A new algorithm, Scalable MapReduce Random Forest (SMRF), was developed and implemented on a computer cluster or cloud environment.

Main Results:

• The SMRF algorithm demonstrated superior overall performance compared to the standard random forest method.
• SMRF achieved comparable accuracy to traditional methods while significantly improving processing throughput.
• The developed approach achieved an accuracy range of 80% for liver cancer prediction.

Conclusions:

• The SMRF algorithm offers an efficient and scalable solution for analyzing massive datasets in disease prediction.
• Big data analytics, combined with machine learning, can accelerate drug discovery and improve diagnostic accuracy for liver cancer.
• The proposed methodology provides a robust framework for developing new medicines and enhancing medical research performance.