An Online Weighted Bayesian Fuzzy Clustering Method for Large Medical Data Sets.

Area of Science:

• Artificial Intelligence
• Medical Informatics
• Data Science

Background:

• The proliferation of AI and wearable devices generates massive medical datasets exceeding single-machine memory capacity.
• Processing large-scale medical data demands significant computational resources, increasing time and hardware requirements.
• Existing data processing methods struggle with the scale and complexity of modern health informatics.

Purpose of the Study:

• To develop an efficient online clustering framework for large-scale medical datasets.
• To reduce memory consumption and accelerate data processing for health informatics.
• To improve clustering performance on diverse medical data types.

Main Methods:

• A novel online clustering framework is proposed, processing data in blocks.
• Weighting clustering is applied to each data block to derive cluster centers and weights.
• Final cluster centers are computed by aggregating block-specific centers and weights.

Main Results:

• The proposed method significantly reduces processing time compared to traditional approaches.
• Experimental results demonstrate superior clustering performance across standard, cancer, and CT image datasets.
• The framework effectively handles large datasets that cannot be loaded into memory at once.

Conclusions:

• The online clustering framework offers an efficient solution for processing large medical datasets.
• This approach accelerates clustering and minimizes memory footprint without compromising performance.
• The method shows promise for applications in health informatics and medical data analysis.