• Version
• Download 3
• File Size 492.61 KB
• File Count 1
• Create Date 9 November 2025
• Last Updated 9 November 2025

Infrastructure Automation for Continuous Validation and Monitoring of ML Models in Hospitals

 

 

Veerendra Nath Jasthi

veerendranathjasthi@gmail.com

 

 

Abstract— With hospitals and other hospital systems undertaking the deployment of machine learning (ML) models to assist in clinical decision-making and diagnostic performance as well as provide operational efficiency, there is a concern whether such models are expected to remain reliable and fair over time. Nonetheless, in case of inefficient infrastructure, the applied models can suffer obsolescence in terms of data drifting, concept drifting, or environmental dynamics. In this paper, the author suggests an infrastructure automation model of constant ML models verification and monitoring in hospitals. The architecture incorporates CI/CD pipelines, container orchestration, the ingestion of data in real time, monitoring dashboards, and drift detection modules. Automatic retraining triggers and model performance alerts can help hospitals maintain strong ML deployments that will adjust with the changes in clinical data. Experiments based on real hospital data (ICU prediction, diagnostic classification, various risks of readmission) show a better ability to retain the accuracy of the models and less control is provided by manual operations. This paper emphasises the significance of MLOps principles applicable in life-threatening environments such as healthcare. In this context, the monitoring and validation processes are life-threatening.

Keywords— Infrastructure Automation, MLOps, Continuous Monitoring, Model Validation, Healthcare AI, Data Drift, Hospitals, Machine Learning, ML Lifecycle, Clinical Decision Support.

Download