Clinical Workflow Automation using ResNet50 based Diagnostic Intelligence

Abstract

This paper presents MediSmart, a web-based hospital management system that automates clinical workflows with multi-role access and smart medical report analysis. The system has four main user roles Patient, Doctor, Receptionist, and Tester with exclusive dashboards for role-based, secure access. Patients book online appointments, which receptionists schedule. Doctors can prescribe or refer diagnostic tests, which testers upload on completion. The most important innovation of MediSmart is the combination of two AI-augmented modules: a rule-based engine for catching unusual values in structured test reports (e.g., blood or ECG readings) and a ResNet50 classifier to identify multi-class cancers from medical images. Both modules produce a concise AI-augmented summary with confidence scores, which is shown to the doctor for analysis, thereby keeping clinical decisions under expert control. The system is developed on the Django web framework (Python) with MySQL database and implements NLP and deep learning methods to carry out end-to-end test report analysis. By eliminating manual effort through automated appointment scheduling, test coordination, and AI-augmented diagnosis, MediSmart increases clinical productivity, eliminates manual effort, and makes data-driven health decisions simpler.

Ähnliche Arbeiten