Future forecasting prediction of Covid-19 using hybrid deep learning algorithm.

Area of Science:

• Medical Informatics
• Artificial Intelligence
• Computational Biology

Background:

• The rapid global spread of COVID-19 necessitates accurate identification and prediction of disease outcomes.
• Existing methods face challenges in real-time analysis and prediction accuracy for coronavirus disease 2019.

Purpose of the Study:

• To develop and evaluate an AI-based deep learning algorithm for detecting COVID-19 positive cases.
• To predict mortality and recovery rates in COVID-19 patients using real-world datasets.
• To enhance the accuracy of coronavirus identification and mortality forecasting.

Main Methods:

• Data pre-processing techniques were applied to clean text datasets.
• Feature extraction was performed using Term Frequency-Inverse Term Frequency (TF-IDF) and Bag of Words (BoW).
• The Mayfly Optimization (MO) algorithm selected relevant features, followed by a hybrid ResNet-GoogleNet deep learning model for prediction.

Main Results:

• The proposed hybrid deep learning method achieved 97.56% accuracy on the first dataset.
• This accuracy surpassed Linear Regression, Multinomial Naive Bayesian, Random Forest, Stochastic Gradient Boosting, Decision Tree, and Bagging techniques.
• The model demonstrated superior performance in identifying COVID-19 and predicting mortality rates compared to existing machine learning models.

Conclusions:

• The developed hybrid deep learning model is effective for COVID-19 identification.
• The AI-based approach provides valuable insights for future mortality forecasting in coronavirus pandemics.
• This research highlights the potential of advanced AI techniques in managing public health crises.