Machine Learning Operations (MLOps) in Healthcare

The operationalization of Machine Learning (ML) models into production environments, a process known as Machine Learning Operations (MLOps), represents a crucial advancement in artificial intelligence. This thesis proposes a novel, standardized methodology composed of ten steps to streamline the deployment of ML models into production. Drawing on a synthesis of existing MLOps practices and frameworks, our study emphasizes integrating continuous integration, delivery, training, monitoring, explainability, and sustainability considerations into the ML lifecycle. We explore the significance of each step, from understanding business problems and data acquisition to model training/testing, deployment, and monitoring, including the pivotal role of explainability and sustainability in fostering trust and reducing environmental impact. Our methodology is validated through practical application in diverse industrial contexts, demonstrating its versatility and effectiveness in enhancing model reliability, efficiency, and alignment with business objectives. This approach bridges the gap between theoretical research and practical implementation. It sets a foundation for future innovations in MLOps, advocating for a more systematic, transparent, and environmentally conscious deployment of ML models in real-world settings.