Google releases MedGemma 1.5 with advanced imaging and MedASR for medical speech-to-text

Google Updates MedGemma With Imaging Support, Launches MedASR for Clinical Dictation | AIM

Both models are available via the AI community Hugging Face and Google Cloud's development platform Vertex AI. Google has released MedGemma 1.5, an updated version of its open healthcare-focused AI model. The company said the model expands support for CT scans, MRI and histopathology, and also introduced MedASR, a medical speech-to-text system for clinical dictation and healthcare workflows. Both models are available via the AI community Hugging Face and Google Cloud's AI development platform Vertex AI. MedGemma 1.5 4B is part of Google's Health AI Developer Foundations programme and is available for research and commercial use. "We are updating our open MedGemma model with improved medical imaging support," said Google in a blog post. "We also describe MedASR, our new open medical speech-to-text model." According to Google, MedGemma 1.5 improved baseline accuracy by 3% on disease classification tasks using CT scans and by 14% on MRI-based classification compared with the previous version. The company also reported gains in anatomical localisation in chest X-rays and structured data extraction from laboratory reports. The 4B-parameter model is designed to be compute-efficient and capable of running offline, while a larger 27B-parameter version remains available for text-heavy medical applications. MedGemma models can be deployed on Google Cloud through Vertex AI, the company said. Google also announced MedASR, an open automated speech recognition model fine-tuned for medical dictation. In internal tests, MedASR recorded a 5.2% word error rate on chest X-ray dictations, compared with 12.5% for a general-purpose speech recognition model. "While text is currently the primary interface for large language models, verbal communication remains crucial in many aspects of healthcare, including medical dictation and live conversations between patients and providers," the company said. Google also announced the MedGemma Impact Challenge, a hackathon hosted on the data science and machine learning platform Kaggle, with $100,000 in prizes to encourage developers to build healthcare applications using the models. The company said MedGemma is intended as a developer foundation model and should be validated and adapted before use in clinical settings. The launch comes amid intensifying competition in healthcare AI. OpenAI recently introduced ChatGPT Health, while Anthropic has rolled out Claude for Healthcare, pushing generative AI deeper into medical workflows.

Unlike OpenAI, Google's Healthcare Push Takes an Open-Source Approach

Google's open-source models are available for research and commercial use Google has introduced two new healthcare-focused artificial intelligence (AI) models, MedGemma 1.5 and MedASR, aimed at improving how medical images and clinical speech data are processed. The release of the open-source AI models marks the next step in the Mountain View-based tech giant's push in the healthcare space. Interestingly, unlike OpenAI, which is offering its ChatGPT for Healthcare as a commercial product for enterprise, the Gemini-maker has taken a community-focused approach by making MedGemma 1.5 and MedASR publicly available. Google Releases MedGemma 1.5 and MedASR AI Models In a blog post, Google Research detailed the new AI models and their capabilities. MedGemma 1.5 is the latest version of Google's open medical vision-language model. It is designed to analyse medical images alongside text, allowing it to interpret scans, answer questions about visual medical data, and support downstream research tasks. The updated version improves on earlier iterations by offering stronger multimodal reasoning, better handling of complex medical imagery, and increased flexibility for fine-tuning on specialised datasets. Google said MedGemma 1.5 can work with different types of medical images, including radiology scans and other clinically relevant visuals. The model is intended to support research use cases such as image-based question answering, report generation, and structured data extraction. The company maintained that the model is not designed to provide diagnoses or treatment recommendations and should be used as a supporting tool in research and development environments. Alongside MedGemma 1.5, Google also introduced MedASR, a medical automatic speech recognition model tailored for healthcare settings. MedASR is designed to convert spoken clinical conversations into text, with a focus on handling medical terminology, accents, and real-world clinical audio conditions. Google said the model aims to reduce errors commonly seen in general-purpose speech recognition systems when applied to healthcare use cases. The company noted that MedASR can be used for tasks such as transcribing doctor-patient interactions, clinical notes, and dictated reports. It is designed to be adaptable across different healthcare environments and can be fine-tuned for specific clinical workflows or documentation standards. Google said that all variants of MedGemma and the MedASR model can be accessed via the company's Hugging Face listing or the Vertex AI platform. Additionally, the tech giant's MedGemma GitHub repository is also available for developers who want to check out the tutorials. Both models come with a permissive licence allowing both research and commercial use cases.

MedGemma 1.5 & MedASR explained: High-dimensional imaging and medical speech-to-text

Google Health AI releases open models for imaging and clinical dictation The adoption of artificial intelligence in healthcare is accelerating at twice the rate of the broader economy, driven by the need for tools that can handle the complexity of medical data. To support this transformation, Google Research has released major updates to its open medical : MedGemma 1.5, a multimodal model with advanced imaging capabilities, and MedASR, a specialized speech-to-text model designed for the medical domain. Together, these models represent a shift from analyzing static, two-dimensional data to interpreting the high-dimensional, multimodal reality of clinical practice. Also read: Optical Illusions: Why AI models are succumbing to the same tricks we do MedGemma 1.5: Seeing in 3D and Time While the original MedGemma 1 could interpret standard 2D images like chest X-rays, medical diagnostics often rely on more complex data. MedGemma 1.5 (specifically the 4B parameter version) bridges this gap by introducing support for high-dimensional imaging. This update allows the model to interpret three-dimensional volume representations, such as Computed Tomography (CT) scans and Magnetic Resonance Imaging (MRI), as well as whole-slide histopathology imaging. Rather than looking at a single slice, developers can now build applications where the model analyzes multiple slices or patches to understand the full context of a scan. Internal benchmarks show significant gains, including a 14% improvement in classifying disease-related MRI findings compared to the previous version. Beyond 3D imaging, MedGemma 1.5 improves longitudinal analysis. In clinical settings, a patient's trajectory is often more important than a single snapshot. The new model excels at reviewing chest X-ray time series, allowing it to track changes over time with greater accuracy. Also read: Anthropic unveils medical AI suite to rival ChatGPT Health: Key features explained The model also boasts improved text-based reasoning, achieving a 90% score on the EHRQA benchmark (Electronic Health Record Question Answering), a 22% jump over MedGemma 1. This ensures that the model is as effective at parsing complex lab reports and medical records as it is at scanning images. MedASR: The Listener Medical documentation relies heavily on dictation, yet general-purpose speech models often stumble over complex medical terminology. MedASR addresses this by providing an open Automated Speech Recognition (ASR) model fine-tuned specifically for medical dictation. When compared to generalist models like Whisper (large-v3), MedASR demonstrated a drastic reduction in errors. It recorded 58% fewer errors on chest X-ray dictations and 82% fewer errors on diverse medical dictation benchmarks. For developers, MedASR is designed to work in tandem with MedGemma. A clinician could potentially dictate a query (processed by MedASR) regarding a specific CT volume (analyzed by MedGemma 1.5), creating a seamless multimodal workflow that mirrors natural clinical interaction. An open foundation for developers Both models are part of Google's Health AI Developer Foundations (HAI-DEF). They are released as open weights, allowing researchers and developers to fine-tune and adapt them for specific use cases - from anatomical localization to structuring data from messy lab reports. By providing these compute-efficient models via platforms like Hugging Face and Vertex AI, Google aims to lower the barrier to entry for creating next-generation medical applications that can see, listen, and reason more like a clinician.