Medikosh — Intelligent Healthcare

Healthcare applications are difficult to build and maintain because they combine sensitive patient data, authentication, file uploads, appointments, external services, and AI-driven clinical workflows. Small software defects can lead to security vulnerabilities, data-isolation problems, unreliable workflows, and significant developer effort during debugging and maintenance. Medikosh is a healthcare platform designed to connect patients, doctors, clinical intake, medical records, appointments, and doctor allocation in one workflow. We enhanced Medikosh with IBM Bob 2.0 as an AI development partner that works with the repository context to improve how the application is analyzed, secured, tested, debugged, and maintained. The platform supports patients in logging in, managing appointments, uploading medical records, and providing clinical information through intake workflows. Doctors can review patient information and manage healthcare workflows. Behind these experiences, Medikosh processes information through authentication, validation, clinical structuring, and doctor-allocation services. Our key innovation is applying IBM Bob not simply as a code generator, but as a repository-aware engineering assistant. Bob can analyze interconnected modules, identify security and maintenance issues, assist with implementing fixes, generate and improve automated tests, and review the resulting changes. During development, this approach helped address real issues such as fake authentication, hardcoded patient identities, unauthorized access risks, insecure CORS configuration, unsafe file uploads, hardcoded password behavior, incorrect appointment-time calculations, and patient-data isolation problems. The target users are patients and healthcare professionals, while developers benefit from the improved engineering workflow behind the platform. The solution follows a validation-first approach: Patient Input → Extraction → Validation → Structured Information → Doctor Review AI-generated information is not treated as automatically trusted clinical information. The system distinguishes between extracted, unverified, validated, and doctor-reviewed information where applicable. The project demonstrates a practical workflow for developing safer healthcare software with AI assistance: Existing Codebase → IBM Bob Analysis → Problem Discovery → Developer Review → Implementation → Automated Testing → Code Review → Verified Changes This makes IBM Bob part of the engineering process rather than an isolated coding tool. The result is a healthcare application with stronger authentication, authorization, data isolation, input validation, testing, and maintainability, while demonstrating how repository-aware AI assistance can reduce manual effort and improve software development workflows.