2025 Reports

Reducing Avoidable Readmissions Through AI-as-a-Service (AIaaS): A Socio-Technical Integration on Google Cloud for Clinical Decision Support

Lamria, Glory

This study proposes a human-centered AI-as-a-Service (AIaaS) solution deployed on Google Cloud to reduce avoidable hospital readmissions in the United States. Leveraging near-real-time electronic health record (EHR) data through FHIR-compliant APIs, the system integrates BigQuery ML and Vertex AI to predict 30-day readmission risk. The architecture enables automated alerts and clinician-facing dashboards via Looker, facilitating proactive intervention within clinical workflows. Beyond technical implementation, the proposal emphasizes a socio-technical approach that integrates cloud infrastructure, network efficiency, and human-centered design to enhance decision-making. The solution targets measurable outcomes, including a 12 percent reduction in readmission rates and significant cost savings, while incorporating ethical safeguards such as bias mitigation, model monitoring, and data security. This work demonstrates how scalable AI systems can shift healthcare delivery from reactive to anticipatory care.

Keywords: AI-as-a-Service (AIaaS), Hospital Readmission, Clinical Decision Support, Electronic Health Records (EHR), Google Cloud Healthcare API, BigQuery ML, Vertex AI, Predictive Analytics, Healthcare AI Ethics, Socio-technical Systems