A Cloud-Integrated Decision Tree Framework for Predictive Modeling in Modern Healthcare

Abstract

The rapid digitization of medical records and the increasing demand for remote patient monitoring have placed unprecedented strain on traditional healthcare IT infrastructures. While predictive analytics offers a pathway to proactive medicine, on-premise systems often struggle with the computational overhead and high latency required for real-time clinical intervention. This paper addresses these limitations by proposing a robust, scalable cloud-based predictive framework centered on a decision-tree model for patient diagnosis and clinical resource optimization in virtual healthcare environments. The proposed system utilizes a multi-cloud strategy to balance specialized workload requirements. AWS SageMaker is employed for the high-performance training and tuning of the decision-tree algorithms, while the Google Cloud Healthcare API facilitates seamless, FHIR-compliant data integration across disparate medical data sources. To ensure the integrity of sensitive patient information, the architecture implements a hybrid cloud deployment model featuring AES-256 end-to-end encryption and strict IAM protocols, meeting global data security standards. Experimental results demonstrate that this cloud-native approach achieves a 95% accuracy rate in predicting disease outcomes, significantly outperforming baseline heuristic models. Furthermore, the transition to cloud infrastructure resulted in a 30% reduction in total computational costs compared to traditional on-premise configurations. Performance benchmarks indicate a high-responsiveness threshold with a latency of ≤200ms, proving the model's viability for real-time decision-making in critical care and virtual triage scenarios. This study underscores the transformative potential of integrating specialized cloud services to enhance the precision and economic efficiency of modern healthcare delivery.