AI-Driven Laboratory Website for Patient Services and Disease Classification Using CNN
Built an AI-powered laboratory platform that streamlines patient services and uses deep learning (CNN) for the automatic detection of pneumonia and lung cancer, developed with PyTorch and TensorFlow.
Visit website
Bringing it together
Laboratories often struggle with managing patient records efficiently while also integrating intelligent diagnostic tools. Before starting this project, patient data and diagnostic workflows were handled separately, leading to delays and limited automation in disease analysis. Our solution was to design a web-based laboratory system that unifies patient management with AI-powered disease classification. Using Convolutional Neural Networks (CNN) built with PyTorch and TensorFlow, the system automatically detects pneumonia and lung cancer from medical images, streamlining both administrative and diagnostic processes.
Our solution was to allow users to be invited to a layer, where they can see others’ annotations and make their own.
Improving the experience
Patients often face long waits, limited access to their results, and complicated payment processes in traditional laboratories. Before this project, there was no unified digital platform for scheduling appointments, managing payments, and accessing medical scans. Our solution was to develop an interactive laboratory website where patients can log in securely, schedule tests, and pay online through BaridiMob or Dahabia. They can also view and download their scans and diagnostic results directly, improving accessibility, convenience, and trust in laboratory services.
Meaningful details
At the core of the system lies an AI-driven disease classification model built using Convolutional Neural Networks (CNN). Two architectures were developed — a binary model for distinguishing between normal and pneumonia cases, and a multi-class model for detecting normal, pneumonia, and lung cancer conditions. Both models were trained on carefully preprocessed medical image datasets to ensure accuracy and reliability. By experimenting with complex CNN architectures, layer optimization, and advanced activation functions, the system achieved strong diagnostic performance, demonstrating how deep learning can enhance medical decision-making and streamline laboratory analysis.
Project outcomes
The developed models achieved high accuracy in classifying medical images across both binary and multi-class datasets, confirming the system’s potential for reliable disease detection. Beyond its technical performance, the project highlights how AI can support diagnostic efficiency and enhance patient care. I aim to collaborate with medical institutions and research professionals to further refine the models, expand dataset diversity, and explore real-world integration for smarter, data-driven healthcare solutions.
