Using Generative AI for Patient Handoff in Healthcare

At the end of each hospital shift, the outgoing nurse has the responsibility of briefing the incoming nurse about the patients under their care. This handoff is a crucial and risky part of the healthcare journey, as accurate information transfer is essential to avoid any lapses in patient care. To address this issue, HCA Healthcare, one of the largest healthcare systems in the US, turned to generative artificial intelligence (AI) solutions. HCA partnered with Google Cloud and utilized their software suite, Vertex AI, to build and deploy machine learning models. While Google offers its own foundation model, PaLM, HCA also explored other options like OpenAI’s GPT-4. In September, Google plans to release a healthcare-specific large language model called Med-PaLM 2 to a wider customer base. HCA, along with other healthcare customers, had early access to this model. However, Microsoft and Amazon are also making their mark in the healthcare AI landscape. Microsoft’s acquisition of Nuance Communications and Amazon’s investments in cloud services showcase their interest in this sector. The competition among these tech giants is still ongoing, and it remains to be seen who will dominate the healthcare AI market. Currently, AI solutions in healthcare focus on solving specific problems, such as aiding in image analysis and speeding up insurance processes. The challenge lies in finding the right large language model and tailoring it to the specific use cases in healthcare. Google recognizes this and is working towards domain adaptation and rigorous quality evaluation of their models. Transparency and constantly updating the training data are concerns in healthcare AI. Google is experimenting with niche models that are trained on narrower sets of data to address these issues. Despite the challenges, HCA sees great potential in AI tools for reducing administrative burdens. They are exploring AI applications for automating medical note-taking and discharge summaries in the emergency room. The consent and privacy of patient data are critically important when using AI in healthcare, and HCA ensures that patient data is used responsibly and compliant with privacy laws. Although there are still hurdles to overcome, HCA remains optimistic that AI tools will become an integral part of medical practice, with doctors leading the way in defining their use for more complex applications. The future of healthcare AI is still being shaped, and tech giants like Google, Microsoft, and Amazon are vying for dominance in this evolving field.