Biotech
Wednesday, December 13th, 2023 3:15 pm EDT
Key Points
- Google’s MedLM Suite Introduction: Google has unveiled the MedLM suite, a collection of healthcare-specific artificial intelligence (AI) models designed to assist clinicians and researchers in conducting complex studies and summarizing doctor-patient interactions. This move is part of Google’s strategy to monetize AI tools in the healthcare industry amid intense competition from rivals such as Amazon and Microsoft.
- Diverse AI Models for Varied Tasks: The MedLM suite includes both a large and a medium-sized AI model, both built on the Med-PaLM 2, a large language model trained on medical data introduced by Google in March. Google recognizes that different medically tuned AI models are better suited for specific tasks. The larger MedLM model is designed for complex tasks requiring deep knowledge and substantial computing power, such as extensive studies using healthcare organizations’ patient population data. Conversely, the medium-sized model is more agile and optimized for specific or real-time functions, like summarizing interactions between doctors and patients.
- Real-world Use Cases and Challenges: Companies like HCA Healthcare and BenchSci have tested Google’s AI technology in real-world scenarios. HCA Healthcare has been using MedLM to automate the documentation of emergency medicine physicians’ interactions with patients, saving significant time and effort. However, challenges include the potential for AI models to generate incorrect information, token limits restricting data feeding to the model, and ongoing efforts to manage the AI over time. Despite the challenges, initial responses from healthcare providers have been positive, emphasizing the need for cautious implementation to avoid risks to patients. Google also plans to introduce healthcare-specific versions of its Gemini AI model into MedLM in the future, aiming to extend its capabilities.
Google has introduced MedLM, a suite of healthcare-specific artificial intelligence (AI) models designed to assist clinicians and researchers in complex studies and doctor-patient interaction summaries. This move reflects Google’s ongoing efforts to monetize AI tools in the healthcare industry amid fierce competition from rivals like Amazon and Microsoft. The MedLM suite comprises large and medium-sized AI models, both built on Med-PaLM 2, a large language model trained on medical data introduced by Google in March. These models are aimed at addressing various healthcare tasks, with the larger model suitable for complex, data-intensive studies, and the medium-sized model optimized for specific or real-time functions like summarizing doctor-patient interactions.
The suite is generally available to eligible Google Cloud customers in the U.S. as of Wednesday, with varying costs based on model usage. Google also plans to incorporate health-care-specific versions of Gemini, its newest and most capable AI model, into MedLM in the future.
Google’s approach involves offering a suite of AI models rather than a one-size-fits-all solution, recognizing that different models excel in specific tasks. The larger MedLM model is geared towards tasks requiring deep knowledge and substantial computing power, such as conducting extensive studies using healthcare organization patient data. In contrast, the medium-sized model is designed for more agile, real-time functions.
Real-world use cases of Google’s AI technology have been tested by companies like HCA Healthcare, one of the largest U.S. health systems, and BenchSci, a company focused on AI in drug discovery. HCA Healthcare has been leveraging Google’s MedLM suite to automatically document emergency medicine physicians’ interactions with patients, thereby streamlining documentation processes. Similarly, BenchSci has merged MedLM’s AI with its own technology to assist scientists in identifying biomarkers crucial for understanding disease progression.
Despite the positive impact observed in these use cases, challenges exist, particularly in the potential for AI models to produce incorrect information. Managing token limits and addressing misinformation generated by AI models remain areas of concern. However, healthcare organizations like HCA Healthcare emphasize the cautious implementation of AI technology to avoid risks to patients.
Google plans to introduce healthcare-specific versions of Gemini into MedLM, extending its capabilities. While Google’s Gemini AI model generated positive interest, concerns arose over a demonstration video not conducted in real time. The company emphasized that the video illustrates possibilities based on real multimodal prompts and outputs from testing.
Google’s MedLM technology is positioned to address specific pain points and challenges within healthcare workflows, offering potential benefits in areas like emergency room documentation and nurse handoff processes. As Google continues to develop and refine its AI models, healthcare organizations are cautiously exploring and implementing these technologies to enhance efficiency and accessibility, recognizing the importance of responsible implementation to ensure patient safety.
For the full original article on CNBC, please click here: https://www.cnbc.com/2023/12/13/how-doctors-are-using-googles-new-ai-models-for-health-care.html
