AI-based lung image analysis enhances disease severity assessment, reducing ICU overload with standardized admission criteria for COVID-19 patients. Expanded AI research is vital for integrating it into clinical practice and preparing for future pandemics. Automated deep learning for COVID-19 lung segmentation and quantification shows potential. However, disparities exist between clinicians' and AI communities' studies on COVID-19 patient care. Integrating AI into clinical practice demands addressing challenges in standardized severity classification, lung lesion characterization, multi-modal imaging integration, robust long COVID quantification, and acute-to-chronic phase understanding. These steps are vital for enhancing patient care. This study aimed to develop a modular AI-based approach for modelling a patient's current state and predicting the short and long-term progression of COVID-19 patients. The specific objectives were to establish a severity assessment system based on the WHO clinical progression scale and to discover the biomarkers that lead to a chronic phase. We developed AssessNet-19, a multi-class radiomics model using chest CTs and standardized WHO-derived severity assessment (Henao et al., 2023). This model improved the accuracy of clinical severity evaluation for COVID-19 patients by 12% compared to radiologists and 11% compared to a single-class lesion model. To achieve this, we curated a diverse, multi-center COVID-19 dataset encompassing radiological, clinical, and laboratory data. This comprehensive dataset reduces biases, enhances generalizability, and includes varied cases, severities, CT scan sources, and contrast use. The AssessNet-19 model with standardized severity scaling and radiological characterization, can be an easy-to-use and versatile foundation for translating findings as clinical decision support systems in hospitals, useful for both the COVID-19 pandemic and future crises.