Artificial intelligence was recently named the most exciting emerging technology in the healthcare industry, and it’s easy to understand why. It has the potential to change healthcare dramatically, from improved administrative processes to drug discovery.
Some of the most profound impacts of AI in healthcare are already happening in clinicians’ offices. With clinical knowledge doubling every 72 days, it can be very difficult for doctors to keep track of updated clinical practice guidelines and keep the latest recommendations top of mind. Generative AI and large language models have the potential to overcome this provider burden. LLMs can ingest great amounts of data from different sources and distill this information into easily understandable insights that doctors can consider when treating patients at the point of care.
However, a recent McKinsey study revealed that many healthcare providers are worried about generative AI risks. For example, there are concerns about exposing patients’ data and creating possible HIPAA violations, and there are questions about the sources of LLM training data. The chance of exposure increases with public LLMs such as ChatGPT, which are not HIPAA-compliant and allow multiple customers to share the same resources.
In response, some healthcare organizations may consider training their own LLMs on patient data, but that’s expensive, time-consuming and requires specialized expertise. This route also puts them at risk of being locked into their models and restricted from trying new and more powerful LLMs. Finally, once a model is trained it can be difficult to discover the source of its recommendations, which could raise reliability questions.
Click the banner below to find out how infrastructure modernization increases healthcare agility.