Healthcare systems increasingly depend on cloud native infrastructures for electronic health records (EHRs), telemedicine platforms, and real-time clinical workflows. Any downtime or performance degradation directly impacts patient care. This paper proposes a predictive maintenance framework that leverages time-series DevOps telemetry logs, metrics, traces, and pipeline events to anticipate failures before they occur. Using cloud monitoring datasets and CI/CD pipeline events, the model performs anomaly detection, resource forecasting, and reliability scoring. Experimental evaluation demonstrates improved up time, reduced MTTR, and proactive mitigation of infrastructure incidents.