A novel bagging C4.5 algorithm based on wrapper feature selection for supporting wise clinical decision making.

Area of Science:

• Medical Informatics
• Machine Learning in Healthcare
• Internet of Things (IoT) in Medicine

Background:

• Effective analysis of long-term health data is crucial for clinical decision-making in Medical IoT systems.
• The high dimensionality and volume of data from Medical IoT healthcare systems present significant challenges for data exploration and management.
• Novel classifiers with effective feature selection are needed to improve classification and prediction performance in healthcare.

Purpose of the Study:

• To propose a novel bagging C4.5 algorithm integrated with wrapper feature selection for clinical decision-making.
• To introduce an innovative sampling method, S-C4.5-SMOTE, to address data distortion and improve system performance.
• To facilitate effective feature selection from large, multi-dimensional medical datasets without computational burden.

Main Methods:

• Development of a novel bagging C4.5 algorithm incorporating wrapper feature selection.
• Introduction of the S-C4.5-SMOTE sampling technique designed to reduce data size without distortion.
• Ensuring dataset balance and smoothness to support effective feature selection.

Main Results:

• The proposed S-C4.5-SMOTE method effectively overcomes data distortion issues.
• The algorithm demonstrates improved overall system performance in handling large medical datasets.
• Successful implementation of wrapper feature selection without the challenge of massive data volumes.

Conclusions:

• The novel bagging C4.5 algorithm with S-C4.5-SMOTE offers a robust solution for analyzing Medical IoT data.
• This approach supports wise clinical decision-making by enabling efficient and accurate data analysis.
• The method represents a significant innovation in managing and extracting insights from high-volume, multi-dimensional healthcare data.