Google has recently announced MedLM, an innovative suite of artificial intelligence models tailored for healthcare applications. These models are specifically designed to assist clinicians and researchers in conducting complex studies, summarising interactions between doctors and patients, and addressing various healthcare-related challenges.
The MedLM suite includes a large and a medium-sized AI model, both of which are built on Med-PaLM 2. This large language model, introduced in March, is trained on medical data, forming the foundation for the capabilities of the MedLM models.
MedLM model is equipped with substantial knowledge and significant computing power to carry out complicated tasks. This includes the ability to conduct studies using data from an entire patient population within a healthcare organisation.
As doctors do not often require help for straightforward inquiries about nature of a disease, the objective of AI in many health organisations is to enhance workflow management. This involves offering more back-office help and addressing perceived slowdowns in the clinicians' daily operations.
For example, Dr. Michael Schlosser, the Senior Vice President of Care Transformation and Innovation at HCA, mentioned the company has been using MedLM to assist emergency medicine physicians in automatically documenting their interactions with patients. Through Google's MedLM suite, transcripts of these interactions are processed to generate the various components of an Emergency Room (ER) provider note.
It is expected that Google’s technology will be able to provide more than half of a provider note without requiring additional input. The goal is for AI to progress to the point where it accurately creates over 60% of the note correctly before human review and editing.
While the response to MedLM has been positive, it's essential to acknowledge that it is still a work in progress. The utmost priority is to ensure responsible implementation of this technology to prevent any potential risks to patients.
Source: Google Cloud
Image Credit: iStock
