Multi-objective deep learning framework for COVID-19 dataset problems.

Area of Science:

• Artificial Intelligence
• Machine Learning
• Medical Informatics

Background:

• The COVID-19 pandemic highlighted challenges in clinical decision-making due to data limitations.
• High missing values in medical datasets hinder accurate patient classification and predictive modeling.

Purpose of the Study:

• To develop and evaluate a robust deep learning framework for handling medical datasets with substantial missing values.
• To improve the accuracy and efficiency of predictive models for diseases like COVID-19.

Main Methods:

• Implemented a Data Missing Care (DMC) framework combined with Grid-Search optimization.
• Assessed three deep learning algorithms: Artificial Neural Network (ANN), Convolutional Neural Network (CNN), and Recurrent Neural Networks (RNN).
• Tuned multiple hyperparameters for optimized predictive model performance.

Main Results:

• Achieved high accuracy (98%), precision (98.5%), F1-score (98.6%), and ROC Curve (95%-99%) on a COVID-19 dataset.
• Demonstrated significant improvement on a second COVID-19 dataset with high missing values, reaching over 91% accuracy.
• Validated effectiveness on a Cervical Cancer dataset, with all metrics exceeding 95%.

Conclusions:

• The proposed hybrid deep learning approach effectively addresses missing data in medical datasets.
• This method offers a high-accuracy, time-efficient alternative to traditional data processing and classification techniques.
• The framework shows potential for broad applications in medicine, energy management, and environmental science.