Google Introduces MedLM: AI Models for Healthcare Tasks Amid Industry Caution

Generative AI meets Healthcare

Google is unveiling its new initiative, MedLM, aiming to leverage generative AI models to enhance healthcare processes. The company believes that these models can assist healthcare professionals in completing tasks more efficiently.

MedLM is built on Med-PaLM 2, a Google-developed model demonstrating "expert level" performance on numerous medical exam questions. Google Cloud customers in the U.S. (with preview availability in select markets) can access MedLM through Vertex AI, Google's fully managed AI development platform, after whitelisting.

There are currently two MedLM models: a larger model for complex tasks and a smaller, fine-tunable model suitable for scaling across various tasks. Google emphasizes the importance of tailoring models to specific use cases, acknowledging that different tasks may require different models.

Early users include HCA Healthcare, a for-profit facility operator, which has been piloting MedLM with physicians to assist in drafting patient notes at emergency department hospital sites. Another tester, BenchSci, has integrated MedLM into its "evidence engine" for identifying, classifying, and ranking novel biomarkers.

While tech giants like Google, Microsoft, and Amazon compete to dominate the potentially lucrative healthcare AI market, there are concerns about the reliability and accuracy of AI in healthcare. Past instances, such as claims made by Babylon Health and the challenges faced by IBM's Watson Health division, highlight the mixed success of AI in the field.

Generative models, like those in Google's MedLM family, are more sophisticated than previous iterations, but studies show challenges in accuracy, particularly in responding to healthcare-related questions. Concerns raised by the World Health Organization (WHO) include the potential for harmful answers, propagation of health misinformation, and the inadvertent leakage of sensitive health data.

Despite these concerns, Google asserts its commitment to cautious implementation, emphasizing a safe and responsible use of generative AI in healthcare. The company aims to contribute to the advancement of healthcare while ensuring that the benefits are accessible to everyone.