Poor and rich dolphin optimization algorithm with modified deep fuzzy clustering for COVID-19 patient analysis.

Area of Science:

• Computational biology
• Data science
• Medical informatics

Background:

• Coronavirus disease 2019 (COVID-19) is a global pandemic causing severe respiratory failure in many patients.
• Understanding disease heterogeneity is crucial for effective patient management and treatment strategies.
• Cluster analysis can identify distinct clinical phenotypes within the COVID-19 patient population.

Purpose of the Study:

• To develop an optimization-driven technique for analyzing COVID-19 patient data.
• To identify and characterize different clinical phenotypes of COVID-19 patients.
• To enhance the accuracy of disease classification and patient stratification.

Main Methods:

• Utilized the Apache Spark framework for distributed data processing.
• Implemented a novel Poor and Rich Dolphin Optimization Algorithm (PRDOA) for feature selection.
• Employed Tanimoto-based Deep Fuzzy Clustering (TDFC) for patient data clustering.
• Combined PRDOA with Tanimoto concept and deep fuzzy clustering for enhanced analysis.

Main Results:

• The proposed PRDOA-TDFC technique demonstrated superior performance in COVID-19 patient analysis.
• Achieved a high clustering accuracy of 89.8%.
• Obtained excellent performance metrics, including Dice coefficient (90%), Jaccard coefficient (85.7%), and Rand coefficient (85.7%).

Conclusions:

• The developed optimization-driven clustering technique effectively identifies COVID-19 patient phenotypes.
• This approach offers a promising tool for improving the analysis and management of COVID-19 patients.
• The findings highlight the potential of advanced computational methods in understanding complex diseases like COVID-19.