Google is expanding its open medical AI offerings with an updated model designed to better interpret complex medical images and a new tool focused on converting medical speech into text.
The company on Jan. 13 released MedGemma 1.5, an updated version of its open medical generative AI model, and introduced MedASR, an open automated speech recognition model tailored for medical dictation, according to a Google Research press release.
MedGemma was first released last year through Google’s Health AI Developer Foundations program as a set of open models intended for developers to adapt for medical use cases. Since its debut, the models have been downloaded millions of times and used to create hundreds of community-built variants, Google said in the release.
The MedGemma 1.5 update focuses on expanding support for medical imaging, particularly higher-dimensional data. The model now supports three-dimensional imaging such as CT scans and MRI volumes, whole-slide histopathology images, longitudinal chest X-ray comparisons and anatomical localization in chest X-rays, according to the release. It also includes improved capabilities for extracting structured data from medical laboratory reports.
In addition to imaging improvements, MedGemma 1.5 includes enhanced medical text performance. Google reported gains on medical question-answering and electronic health record retrieval tasks after adding new training datasets and techniques.
Alongside the MedGemma update, Google introduced MedASR, a speech-to-text model trained specifically for medical language. The company said the model produces significantly fewer transcription errors than a general-purpose speech recognition model on internal benchmarks involving medical dictation. MedASR is designed to support both transcription and voice-based interaction with MedGemma for downstream reasoning tasks, according to the release.
Both MedGemma 1.5 and MedASR are available for research and commercial use through platforms including Hugging Face and Google Cloud’s Vertex AI, Google said.
At the Becker's 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health, taking place September 14–17 in Chicago, healthcare executives and digital leaders from across the country will come together to explore how AI, interoperability, cybersecurity, and revenue cycle innovation are transforming care delivery, strengthening financial performance, and driving the next era of digital health. Apply for complimentary registration now.