Google Announces Med-Gemini AI Models for Healthcare Use
Google has introduced Med-Gemini, a family of multimodal models built upon Gemini specifically designed for the healthcare industry. While the models are still unavailable for public or beta testing, the tech giant has published a detailed pre-print version in its research paper available on arXiv.
Key Features of Med-Gemini AI Models
A notable feature of the AI model is its long-context ability, which enables better processing of health records and research papers. Further, all of the AI models are multimodal and can provide text, image, and video outputs.
Jeff Dean, Chief Scientist at Google DeepMind and Google Research, expressed his excitement about the potential of these models in healthcare, stating, "I'm very excited about the possibilities of these models to help clinicians deliver better care, as well as to help patients better understand their medical conditions. AI for healthcare is going to be one of the most impactful application domains for AI, in my opinion."
Performance and Capabilities
Google claims that Med-Gemini AI models have outperformed OpenAI's GPT-4 models in the GeneTuring dataset on text-based reasoning tasks. The models, including Med-Gemini-S 1.0, Med-Gemini-M 1.0, Med-Gemini-L 1.0, and Med-Gemini-M 1.5, build upon Gemini 1.0 and Gemini 1.5 LLM.
Reportedly, Med-Gemini-L 1.0 has shown a 4.5% improvement over its predecessor Med-PaLM 2, scoring 91.1% accuracy on MedQA (USMLE).
Enhanced Search Integration
Web search integration has enhanced the models, making them "more factually accurate, reliable, and nuanced," which is evident in the results for complex clinical reasoning tasks, according to Google. The AI model is fine-tuned for improved performance during long-context processing, aiming for higher-quality long text processing to provide accurate answers and insights.
Google is focused on further improving the model before making it available to the public.