HClass: Automatic classification tool for health pathologies using artificial intelligence techniques.

Area of Science:

• Medical Informatics
• Machine Learning in Healthcare
• Computational Pathology

Background:

• Accurate classification of patient pathologies is crucial for effective treatment, yet complex variables can lead to diagnostic confusion.
• Existing diagnostic methods may lack the precision needed for nuanced pathological distinctions.
• The integration of advanced computational techniques offers a potential solution to enhance diagnostic accuracy.

Purpose of the Study:

• To develop and evaluate a novel machine learning algorithm (hClass) for the precise classification of subject pathologies.
• To improve the rigor of disease treatment by reducing diagnostic ambiguity in clinical practice.
• To provide a user-friendly tool for classifying health data with high accuracy.

Main Methods:

• Application of machine learning techniques, including supervised, non-supervised, and semi-supervised classification, to a health-record database.
• Configuration of the machine learning model using cross-validation for set validation and Principal Component Analysis (PCA) for feature reduction.
• Implementation of classifier committees for ensemble assessment and performance enhancement.

Main Results:

• The hClass algorithm demonstrated a high success rate in classifying pathologies.
• The ADABoost classifier achieved a 90% success rate.
• A committee of three classifiers, combined with PCA, yielded an 89.7% success rate, indicating robust performance.

Conclusions:

• The hClass algorithm offers a reliable and accurate method for classifying subject pathologies, enhancing diagnostic precision.
• Machine learning, particularly with techniques like PCA and ensemble methods, significantly improves the accuracy of health data classification.
• The developed tool is adaptable and can be further optimized for diverse classification tasks in healthcare.