Hepatocellular Carcinoma Detection Using Machine Learning Techniques.

Area of Science:

• Oncology
• Bioinformatics
• Machine Learning

Background:

• Hepatocellular carcinoma (HCC) is a primary liver cancer.
• Accurate early detection and diagnosis are crucial for patient outcomes.
• Machine learning offers potential for improving diagnostic accuracy.

Purpose of the Study:

• To evaluate various machine learning algorithms for HCC classification.
• To identify the most effective algorithm for HCC detection using a public dataset.
• To develop a decision support system for clinical use.

Main Methods:

• Utilized the hepatocellular carcinoma dataset from the UCI machine learning repository.
• Applied and compared multiple classification algorithms: decision trees, random forests, SVMs, k-NN, AdaBoost, and gradient boost.
• Performed feature selection and classification model evaluation.

Main Results:

• Gradient boost algorithm demonstrated superior performance.
• Achieved 84% accuracy and 93% precision with the gradient boost model.
• Other algorithms showed varying levels of effectiveness.

Conclusions:

• Gradient boost is a highly effective machine learning technique for HCC classification.
• The developed model can serve as a valuable decision support tool for clinicians.
• Further research can explore integration into clinical workflows.