An Integrated Multi-Agent Frameworks for COVID-19 Detection Using Machine Learning and Deep Learning Techniques

The SARSCoV-2 virus causes the infectious illness known as coronavirusdisease (COVID-19). In recent decades, COVID-19 has become the mostinfectious disease. Millions of individuals worldwide suffer from thisdisease. Because of the restricted availability and sensitivity of testingkits, doctors and researchers have turned to computer tomography(CT) scans to screen for COVID-19. Recent technological developmentsand the widespread use of machine learning (ML) and deep learning(DL) techniques have shown high potential in terms of more accurateCOVID-19 detection. Therefore, in this paper, we have developed twomulti-agent frameworks for COVID-19 detection using the ML and DLalgorithms. In the first framework, several ML algorithms, namely, KNN, LR, and SVM, are employed. Further, ensemble learning and hypertun-ing of the ML algorithms is done using the grid search method. Next, reinforcement learning method Q learning agent is used to update themulti-agent framework. On the other hand, the second multi-agentframework is developed with the help of lightweight ResNet andreinforcement PPO algorithms. Further, in both frameworks, featureselection is done using the LRP-ET method to determine the appropriatefeatures from the dataset. The experimental results are performed onthe Google Colab software for the standard dataset. The dataset is splitinto an 80:20 ratio to train and test the frameworks. The evaluationof both frameworks is done using the various parameters, namely,accuracy, precision, recall, and F-score. The results show that bothframeworks outperform the existing models. Finally, the comparisonof both frameworks shows that the second framework, which is basedon deep learning, performs superiorly over the first framework, whichis based on ML, due to efficiently handling complex data. How to cite this article:Walia R K, Kaur H. An Integrated Multi-AgentFrameworks for COVID-19 Detection UsingMachine Learning and Deep Learning Techniques.J Commun Dis. 2025;57(1):20-32. DOI: https://doi.org/10.24321/0019.5138.202503