Published Paper Details:

Cataract is one of the top reasons people lose their eyesight worldwide, yet it can be easily treated if found early. The real problem is that most people -- especially in villages and small towns -- do not have easy access to eye doctors. To help solve this, we built Cataract Hub, a desktop application that uses artificial intelligence to check whether an uploaded eye photograph shows signs of cataract. The system uses five deep learning models -- AlexNet, VGG16, ResNet18, DeepCNN, and DeepANN -- that all examine the same image and vote on the result. The final answer is decided by the majority vote, which makes the system much more reliable than relying on one model alone. Before analysis, the image goes through a seven-layer validation check to make sure it is a proper eye photo. Along with the Cataract or Normal verdict, the system also measures four clinical eye features, provides a three-level explanation for patients, doctors, and data scientists, and generates a full medical report using the LLaMA 3.3-70B large language model through the Groq API. Users can also chat with the built-in Medical AI assistant in English, Telugu, or Hindi. The entire system runs as a native desktop application built with Electron JS, making it practical and easy to install in clinics, hospitals, and health camps.