Google on Tuesday announced a new family of artificial intelligence (AI) models focused on the medical field. These AI models of his, called Med-Gemini, are not available to the general public, but the tech giant has published a preprint version of a research paper highlighting their capabilities and methodology. The company claims that its AI model outperforms his GPT-4 model in benchmark tests. One of the notable features of this particular AI model is its long-context capabilities that allow it to process and analyze health records and research papers.
The research paper is currently in pre-print stage and published on arXiv, an open-access online repository for academic articles. Jeff Dean, Principal Scientist at Google DeepMind and Google Research, said: post X (formerly known as Twitter) is “extremely excited about the potential of these models to help clinicians provide better care and help patients better understand their medical conditions. In my opinion, AI for healthcare will be one of the most impactful application domains for AI.”
Med-Gemini is built on Gemini-1.0/1.5 and can be easily adapted to new medical modalities using custom encoders. We present his Med-Gemini promise in precise multimodal dialogue with examples of quality conversations about radiology and dermatology images 🗣️🖼️. pic.twitter.com/VSdU3fvijk
— Alan Karthikesalingam (@alan_karthi) April 30, 2024
Med-Gemini AI models are built on Gemini 1.0 and Gemini 1.5 LLM. There are four models in total: Med-Gemini-S 1.0, Med-Gemini-M 1.0, Med-Gemini-L 1.0, and Med-Gemini-M 1.5. All models are multimodal and can provide text, image, and video output. The model is also integrated with web search, and the company says the model is “more factually accurate, reliable, and nuanced” when displaying results for complex clinical reasoning tasks. It is said that it was improved through self-training to become.
Additionally, the company claims that the AI model has been fine-tuned to improve performance while processing long contexts. High-quality long-context processing means chatbots can provide more accurate and pinpoint answers even when questions are not fully queried or when long documents from medical records need to be processed. .
According to data shared by Google, the Med-Gemini AI model outperformed OpenAI's GPT-4 model on text-based inference tasks in the GeneTuring dataset. Med-Gemini-L 1.0 also recorded an accuracy of 91.1 percent in MedQA (USMLE), outperforming its own older model Med-PaLM 2 by 4.5 percent. In particular, AI models are not available for public or beta testing. The company is likely to further refine the model before putting it in the public domain.
