# MedGemma

The MedGemma collection contains Google's most capable open models for medical
text and image comprehension, built on Gemma 3. Developers can use MedGemma to
accelerate building healthcare-based AI applications. The following versions of
MedGemma are available:

- MedGemma 1.5: 4B multimodal
- MedGemma 1: 4B multimodal version and 27B text-only and multimodal versions.

For details about how to use the model and how it was trained, see the [MedGemma
model card](https://developers.google.com/health-ai-developer-foundations/medgemma/model-card).

## Common use cases

The following sections present some common use cases for the model. You're free
to pursue any use case, as long as it adheres to the [Health AI Developer
Foundations terms of
use](https://developers.google.com/health-ai-developer-foundations/terms).

### Medical image interpretation

MedGemma's pre-trained multimodal variants are well-suited for tasks like
generating medical image reports or answering natural language questions about
medical images. While its baseline performance is strong compared to similar
models, MedGemma isn't yet clinical-grade and will likely require further
fine-tuning.

MedGemma 1.5 4B enables developers to more effectively adapt MedGemma for
applications that involve several medical imaging modalities including
high-dimensional medical imaging CT (Computed Tomography), MRI (Magnetic
Resonance Imaging), Whole-slide histopathology imaging (WSI), longitudinal
medical imaging, and anatomical localization:

### Medical text comprehension and clinical reasoning

MedGemma can be adapted for use cases that require medical knowledge. Such use
cases may include patient interviewing, triaging, clinical decision support, and
summarization. For most use cases, the larger MedGemma 27B model will generally
yield the best performance. Both sizes of MedGemma have a strong baseline
performance compared to similar models of their size, but developers should
validate their adapted model's performance and make necessary improvements
before deploying in a production environment.

MedGemma 1.5 4B delivers improved accuracy on medical text reasoning. The
updated model specifically expands support for data processing applications
including medical document understanding via extraction of structured data from
unstructured medical lab reports and Electronic Health Record (EHR)
understanding for the interpretation of text-based EHR data.

### Adapting MedGemma

MedGemma is a developer model that requires validation on the developer's
intended use case. Based on those validation results, the user will likely need
to further adapt the model to improve performance. Below are some types of
adaptation developers can use to improve MedGemma's performance for their use
cases.

### Prompt engineering/in-context learning

For certain use cases, MedGemma's baseline performance may be sufficient after
careful prompting, potentially including few-shot examples of desirable example
responses within the prompt, in other words in-context learning. Prompt
engineering may also use MedGemma to break the task into subtasks that can be
performed separately. Adaptations using prompt engineering require the same
level of validation as any other type of adaptation.

### Fine-tuning

MedGemma can be fine-tuned for improved performance on the existing tasks it's
been trained on, or to add additional tasks to its repertoire. For an example of
how to fine-tune MedGemma using [LoRA](https://arxiv.org/abs/2106.09685) (a
parameter-efficient fine-tuning technique), see this
[notebook](https://colab.research.google.com/github/google-health/medgemma/blob/main/notebooks/fine_tune_with_hugging_face.ipynb).
To see an example of the use of reinforcement learning with MedGemma, see this
[notebook](https://github.com/Google-Health/medgemma/blob/main/notebooks/reinforcement_learning_with_hugging_face.ipynb).

Users can specifically fine-tune the language model decoder component to help
the model better interpret the visual tokens produced by the image encoder, or
fine-tune both.

### Agentic orchestration

MedGemma can be used as a tool within an agentic system, coupled with other
tools, like web search, [FHIR](https://www.hl7.org/fhir/overview.html)
generators/interpreters, [Gemini
Live](https://gemini.google/overview/gemini-live) for bidirectional audio
conversation, or [Gemini 2.5
Pro](https://deepmind.google/technologies/gemini/pro/) for function calling or
reasoning. MedGemma can also be used to parse private health data locally before
sending anonymized requests to centralized models like Gemini 2.5 Pro.

## Next steps

- [Get started using the model](https://developers.google.com/health-ai-developer-foundations/medgemma/get-started)