HaomaX is a platform that automatically generates radiology reports. The idea was sparked by a brief conversation with a tele-radiologist at a social event in Manhattan, NY (winter 2023). During our discussion, I learned that most radiology reports follow structured patterns, making them ideal candidates for automation using recent advances in Generative AI.

At the time, LLaVa2 had just been released. As the first open-source multimodal LLM capable of integrating visual features from images with textual inputs to produce highly accurate outputs, it was the perfect tool for this challenge. Leveraging LLaVa2 and publicly available X-ray datasets, I developed a platform that enables healthcare providers to upload medical images and receive comprehensive reports in under a minute.

This project not only highlights the transformative power of GenAI in healthcare but also demonstrates how cutting-edge LLMs can address real-world challenges with efficiency and precision.

[Link to our registered invention]