Google Health Developing GenAI Models for Fitbit, Clinicians

Google Health has announced some of the latest strides it has made in the healthcare industry with its use of generative artificial intelligence (GenAI) models.

Last year, the company introduced Med-PaLM 2, a large language model (LLM) specifically fine-tuned for healthcare, Yossi Matias, vice president of engineering and research at Google, wrote in a Tuesday (March 19) blog post. This model has since become available to global customers and partner organizations, enabling them to streamline nurse handoffs and support clinicians’ documentation.

Building on the success of Med-PaLM 2, Google Health introduced MedLM at the end of 2023. This is a family of foundation models for healthcare offering broader availability through Google Cloud’s Vertex AI platform, according to the post.

Google Health’s models for healthcare incorporate new modalities, such as radiology images, lab results, genomics data and environmental context, the post said. One such model, MedLM for Chest X-ray, has the potential to revolutionize radiology workflows by aiding in the classification of chest X-rays. This technology is currently available to trusted testers in an experimental preview on Google Cloud.

Google Health is also investing in research to fine-tune its models for the medical domain, per the post. With approximately 30% of the world’s data volume being generated by the healthcare industry, there is a wealth of information to leverage. However, relevant information is often buried deep within medical records.

To address this issue, Google Health is researching how a version of the Gemini model, fine-tuned for the medical domain, can unlock advanced reasoning capabilities and process multiple modalities, according to the post. This research has already achieved state-of-the-art performance on benchmarks like the U.S. Medical Licensing Exam (USMLE)-style questions and the MedVidQA video dataset.

In collaboration with Google Research and Fitbit, Google Health is developing a Personal Health Large Language Model, the post said. This model aims to power personalized health and wellness features in the Fitbit mobile app, providing individuals with tailored coaching and recommendations based on their health and fitness goals. 

Generative AI is already proving to be an assistive tool for clinicians, helping them with administrative tasks such as documentation, per the post. Google Health is expanding on this work by collaborating with partners to explore further possibilities. One notable development is the introduction of AMIE (Articulate Medical Intelligence Explorer), a research AI system optimized for diagnostic reasoning and clinical conversations. 

PYMNTS Intelligence has found that GenAI will play a critical role in the transformation of the healthcare industry. Potential applications of the technology include improvements in drug discovery, diagnostics and patient care, according to “Generative AI Can Elevate Health and Revolutionize Healthcare,” a PYMNTS Intelligence and AI-ID collaboration.