AI Policies and Regulatory Challenges

According to Stat Healthcheck, the regulatory challenges posed by large language models are verification, provenance, changes, usability, risks and surveillance.  More details are provided in the chart below.

By now, it’s clear that while large language model technologies similar to OpenAI’s ChatGPT might be useful for some health care tasks, like drafting clinician emails, their propensity to fabulation make them a risky choice for diagnostic tools and other medical implementations. Authors of a new article in Nature Medicine dive into another challenge: Many use cases, like tools that help doctors make decisions, may require regulatory clearance in both the United States and Europe — which could prove very difficult because of their “inherent limitations” (see above). The authors suggest approaches that could make regulation possible, including making the underlying algorithms and data models available for review or introducing oversight tools that keep tabs on all chats independent of the developer. That sounds like a lot for companies that closely guard their secret technologies, but the authors insist that “the huge efforts applied to training LLMs for creativity and plausibility should equally be applied to providing evidence of safety and validation in medicine.”

Michael LeTan at Healthcare Risk Advisors notes that healthcare professionals and hospitals are quickly implementing guidelines, policies and even restrictions, cautioning health professionals about the accuracy of AI-generated information, and how it can be used in clinical settings.  One practise is to restrict the input of privacy protected information, information of commercial value (e.g. software) or a person or institutions’ proprietary or intellectual property.

In a recent Globe and Mail article, Deb Pimental, Vice-President of Data, AI and Automation at IBM Canada, said, “Any organization that wants to apply an AI model needs to establish guardrails to ensure that AI is transparent, governed and trusted. That’s at the core of the work we do. At IBM, we’ve got ethical practices and design principles that augment people, not replace them, so we’re looking to free them up to do much more interesting work.”

For a guide on implementing Generative AI in healthcare, check out Health AI partnerships.  They “… empower health care professionals to use AI effectively, safely, and equitably through community-informed up-to-date standards.”

Organizations, especially in healthcare, need to have strong data governance strategies.  This is key to understanding where the data collected by the LLM is from, how it has been performing historically, including its accuracy, what biases are inherent in the data collection, and what controls are in place for who uses it and when.

However AI is used, it is critical to document how it was used.

 

For more articles in the series on Artificial Intelligence, go to:

2Ascribe Inc. is a medical and dental transcription services agency located in Toronto, Ontario Canada, providing medical transcription services to physicians, specialists (including psychiatry, pain and IMEs), dentists, dental specialties, clinics and other healthcare providers across Canada. Our medical and dental transcriptionists take pride in the quality of your transcribed documents. WEBshuttle is our client interface portal for document management. 2Ascribe continues to implement and develop technology to assist and improve the transcription process for physicians, dentists and other healthcare providers, including AUTOfax. AUTOfax works within WEBshuttle to automatically send faxes to referring physicians and dentists when a document is e-signed by the healthcare professional. As a service to our clients and the healthcare industry, 2Ascribe offers articles of interest to physicians, dentists and other healthcare professionals, medical transcriptionists, dental transcriptionists and office staff, as well as of general interest. Additional articles may be found at http://www.2ascribe.com.  For more information on Canadian transcription services, dental transcription, medical transcription work or dictation options, please contact us at info@2ascribe.com.

You might also enjoy

AI Bias

The results created by an AI model can be considered