Russian Researchers Build AI App to Spot Lung Diseases
At Penza State University, a team has created “Radex,” an AI-driven assistant that delivers preliminary diagnoses by detecting signs of 18 different lung conditions.
The app identifies pathologies such as pneumonia, pleurisy, pneumothorax, fibrosis, tumors, and more. Doctors upload an X-ray and enter a command in chat, and within 30 seconds the assistant provides a preliminary diagnosis with up to 70% accuracy. Physicians can ask follow-up questions and maintain a dialogue with the system. The tool can also highlight abnormal areas directly on the image.
Radex is powered by multimodal language models for image analysis and neural networks, integrated into a single architecture overseen by a supervisory agent. Each model focuses on a specific task, while the agent coordinates their work, manages decision-making logic, and dynamically selects tools depending on the case.
The researchers have already received state registration for the software and are now refining the web application.
