This paper presents a computer-aided diagnosis (CAD) system designed to assist physicians in identifying COVID-19 through CT imaging analysis. The system, named RACNet, is a state-of-the-art deep learning model that integrates classification and segmentation components to automatically analyze 3D CT scans, providing infection probability and enhancing diagnostic efficiency. The system addresses challenges such as data discrepancy, anonymization, and security, and is deployed on both cloud and edge environments. Key features include effectiveness, data anonymization, fairness, and enhanced explainability. The RACNet model, based on a CNN-RNN architecture, dynamically selects RNN outputs for decision-making, and extracts latent variables to generate anchor sets that provide insights into the network's knowledge. The system's deployment is facilitated by a microservices architecture, MLPod™, which ensures secure and scalable operation. The paper also discusses the ethical standards compliance and future work, including user feedback collection for model retraining.This paper presents a computer-aided diagnosis (CAD) system designed to assist physicians in identifying COVID-19 through CT imaging analysis. The system, named RACNet, is a state-of-the-art deep learning model that integrates classification and segmentation components to automatically analyze 3D CT scans, providing infection probability and enhancing diagnostic efficiency. The system addresses challenges such as data discrepancy, anonymization, and security, and is deployed on both cloud and edge environments. Key features include effectiveness, data anonymization, fairness, and enhanced explainability. The RACNet model, based on a CNN-RNN architecture, dynamically selects RNN outputs for decision-making, and extracts latent variables to generate anchor sets that provide insights into the network's knowledge. The system's deployment is facilitated by a microservices architecture, MLPod™, which ensures secure and scalable operation. The paper also discusses the ethical standards compliance and future work, including user feedback collection for model retraining.