Google has introduced MedLM, a suite of foundational models customised for applications in the healthcare industry.
The company said MedLM is now accessible to Google Cloud users in the US, with a designated general availability on the Vertex AI platform. Additionally, MedLM is presently in a preview phase in select global markets.
The current line-up of MedLM comprises two models, both derived from Med-PaLM 2, providing versatility for diverse healthcare organisational needs.
As healthcare institutions increasingly explore AI applications for a spectrum of tasks, from basic functions to intricate workflows, Google’s experience in piloting these tools has revealed that the most effective model varies based on the specific use case.
For instance, one model may excel in summarising conversations, while another may be better suited for searching through medications.
The initial MedLM model is designed for handling complex tasks, offering a larger capacity, while the second model is of medium size, designed for fine-tuning and scalability across various tasks.
The development of these models has been guided by the distinct requirements of healthcare and life sciences customers, addressing needs such as responding to medical queries from healthcare providers and generating summaries.
Google plans to integrate Gemini-based models into the MedLM suite to further enhance its capabilities in the upcoming months.
Several companies that have tested MedLM are now progressing towards its integration into their production solutions or expanding their testing efforts.
Over the preceding months, HCA Healthcare has been conducting trials of a solution aimed at aiding physicians in managing their medical notes across four emergency department hospital sites.
Physicians leverage an Augmedix app installed on hands-free devices to generate precise and timely medical notes derived from clinician-patient interactions, adhering to the Health Insurance Portability and Accountability Act (HIPAA).
Augmedix’s platform incorporates natural language processing, in conjunction with Google Cloud’s MedLM on Vertex AI, to swiftly transform data into preliminary drafts of medical notes. Subsequently, physicians review and finalize these drafts before they are seamlessly transmitted in real-time to the hospital’s electronic health record (EHR).
The integration of MedLM is poised to enhance the automated functionality of Augmedix’s ambient documentation products, with an ongoing refinement of quality and summarisation capabilities.
Furthermore, the inclusion of MedLM facilitates a cost-effective expansion of Augmedix’s products across health systems, catering to an expanding array of medical subspecialties, including primary care, emergency departments, oncology, and orthopaedics.
In a strategic move, MedLM is being seamlessly integrated into BenchSci’s ASCEND platform, thereby advancing the pace and quality of pre-clinical research and development.
Google has partnered with Accenture to enhance enterprise adoption, aiding healthcare organisations in leveraging MedLM. Additionally, collaborations with Deloitte and healthcare providers were established to assist care teams in extracting information from provider directories and benefits documents.