Google revealed a collection of AI models labeled MedLM on Wednesday. This move reflects Google's attempt to gain financial benefit from artificial intelligence applications in the healthcare business, where it faces rivalry from Amazon and Microsoft. The firm has plans to add health-oriented versions of Gemini, their most cutting-edge AI model, to MedLM in the near future. like summarization of doctor-patient interactions and complex data studies of medical populations have been the focus of Google's latest effort to create health-care-specific AI tools. On Wednesday, the search giant released the MedLM suite, which includes both a large and medium-sized AI model designed to help clinicians and researchers. Companies like HCA Healthcare, which have been running tests with Google's technology, have seen its potential for impact.Experts are taking steps to implement it cautiously, while acknowledging that the competition for market share between Google, Amazon, and Microsoft remains fierce. Google said the cost for the suite varies depending on the model used, but the medium-sized one is the least expensive to run. Additionally, it plans to introduce health-care-specific versions of its newest and most powerful AI model, Gemini, to MedLM in the future.
When Google announced Med-PaLM 2 in March, the company initially suggested that it could be applied to answer queries such as "What are the first warning signs of pneumonia?" and "Can incontinence be cured?" But as the company has tested the technology with customers, the use-cases have shifted, according Greg Corrado, head of Google's health AI. Corrado stated that clinicians don't usually require assistance with “accessible” inquiries regarding an illness, so Google has not observed a lot of interest for those abilities from customers. Instead, health organizations tend to seek AI to help tackle more back-office or logistical issues, such as managing paperwork. "They want something that can attend to the true pain-points and slowdowns that are part of their workflow, which only they know," Corrado told CNBC.
As an example, HCA Healthcare, one of the largest healthcare systems in the US, has been experimenting with Google’s AI technology since the spring. The company declared an official partnership with Google Cloud in August which aims at using its generative AI to “enhance workflows on lengthy tasks”. Dr. Michael Schlosser, senior vice president of care transformation and innovation at HCA, said the company has been using MedLM to aid emergency medicine physicians automatically document their interactions with patients. For instance, HCA has been employing an ambient speech documentation system from a company known as Augmedix to transcribe doctor-patient conversations. Google's MedLM suite can take these transcripts and separate them into the components of an ER provider note.
Schlosser said that HCA has been utilizing MedLM within emergency rooms at four hospitals, and the company wants to expand use over the following year. By January, Schlosser added, he anticipates Google's technology will be able to accurately create more than half of a note without help from providers. With doctors who can spend up to four hours daily on clerical paperwork, Schlosser commented that saving that time and effort makes a meaningful difference. “That has been a huge jump forward for us,” Schlosser told CNBC. “We assume that we will soon be at a point where the AI alone can create 60-plus percent of the note accurately before we have the human conducting the review and the editing.”
Schlosser also said HCA is also making an effort to employ MedLM to develop a handoff tool for nurses. The tool can read through the electronic health record and identify relevant information for nurses to pass on to the following shift. Handoffs being “laborious” and a real problem for nurses, so it would be “strong” to automate the process, Schlosser commented. Nurses across HCA’s hospitals conduct about 400,000 handoffs a week, and two HCA hospitals have been testing the nurse handoff tool. Schlosser said nurses carry out a side-by-side comparison of a traditional handoff and an AI-generated handoff and provide feedback.
With both use cases, though, HCA has noticed that MedLM is not foolproof. Schlosser said the fact that AI models can spew out improper information is a major challenge, and HCA has been working with Google to devise best practices to reduce such fabrications. He added that token limits, which limit the amount of data that can be provided to the model, and monitoring the AI over time have been extra challenges for HCA. “What I would say right now, is that the hype around the current use of these AI models in healthcare is surpassing the reality,” Schlosser said. “Everyone is coming across this problem, and no one has truly let these models run free in a scaled way in the healthcare systems due to that.”
Even so, Schlosser declared that providers’ preliminary response to MedLM has been agreeable, and they recognize that they are not dealing with the completed product yet. He said HCA is striving hard to implement the technology in a responsible manner to prevent putting patients at risk. “We're being very cautious with how we approach these AI models,” he said. “We're not using those use cases where the model outputs can somehow alter someone’s diagnosis and treatment.”
Google's shares rose 5% following the launch of Gemini earlier this month, but the company was met with criticism over its demonstration clip, which it has confirmed to Bloomberg wasn't conducted in real time. Responding to this news, Google told CNBC: "The clip is a representation of the capabilities of Gemini, based on actual multimodal prompts and outcomes from testing. We anticipate exploring what users can accomplish once Gemini Pro is available starting December 13." Corrado and Gupta from Google revealed that Gemini is still at an early stage and must be tested by customers in a controlled health-care environment prior to being deployed via MedLM. According to HCA's Schlosser, the firm is "very excited" about Gemini and is already establishing plans to analyze the technology. BenchSci, a business using AI to resolve issues associated with drug discovery, is another company that has been working with MedLM. Google being one of its investors, BenchSci has tested the MedLM technology for a few months and has conformed it to its own technology, enabling scientists to detect biomarkers, thereby gaining an insight into the progression and treatment of illnesses. Belenzon, BenchSci's co-founder and CEO, declared that the company invested a great deal of time examining the model and providing Google feedback as to what enhancements were needed. Deloitte has also been utilizing MedLM and has been conducting numerous tests to make sure the technology is ready for implementation to health-care customers. This AI-driven system can help health systems and health plans answer queries from members regarding access to care by searching for providers based on gender, vicinity, benefit coverage, or other qualifiers. Despite its accuracy and effectiveness, Gebreyes from Deloitte notes that it can be challenging to make use of the technology if patients are unsure of the correct term or spelling. "Ultimately, this does not substitute a diagnosis from a trained professional," he added. "It brings expertise closer and makes it more accessible."
Comments