Before integrating a new technology into your organization, it’s important that you’ve thought through all of the potential risks and concerns that can come with that technology. So while there are a lot of benefits to come from ambient clinical voice technology, we have to be sure that we are prepared for the challenges. As it is only through being prepared that we can prevent these risks.

So what are the risks and concerns of ambient clinical voice technology and how could they be holding people back from adopting it? In search of an answer to this question, we reached out to our insightful Healthcare IT Today Community and this is what they had to share with us.

Detlef Koll, Vice President Global R&D at 3M Health Information Systems

Individual users will need to be diligent in reviewing draft notes to avoid incorrect or incomplete capture of information. When front-end speech recognition technology was broadly introduced across all specialties 10-15 years ago, at first, speech recognition accuracy tended to be good enough to reduce the administrative effort of clinical documentation, but also low enough to strongly encourage diligent proofreading. Over time technology improved, errors became less common, and users came to expect close to perfect output.

While overall desirable, this can have a detrimental impact on proofreading discipline. The potential negative impacts of this are well documented; quality assurance processes can effectively address this risk to documentation quality. AI-generated output appears plausible even if the content of the note is factually incorrect. As the technology improves over time, there is a risk that documentation errors in AI-generated drafts might get accepted into a completed note.

To address this risk at an enterprise level, physician leadership (CMO, compliance officers) will need to establish that best practices are followed consistently across the physician population. AI solutions must provide the means for scalable and robust quality assurance processes as a precondition for enterprise-wide rollout.

Etienne Boshoff, Managing Director at EHR Enhancify

As Ambient Clinical Voice Technology becomes increasingly prevalent in healthcare settings, there are several critical considerations to be made for privacy and secure handling of patient data. The key one is that this continuous collection and analysis of patient data raises the risk of data leaks and hacking, and so the technology vendor and the organization using it must ensure they have robust security measures in place. In fact, the best solution is for healthcare providers, technology developers, and regulatory bodies to work together to address these considerations and ensure the secure and ethical use of this promising technology.

One potential issue we can foresee is technical error caused by inaccuracies or bugs in the voice recognition technology. It also might struggle with different languages and accents. All of this could lead to incorrect data being recorded, stored, and used for patient care. Without proper oversight and reviews, overreliance on this technology could lead to incorrect prescriptions and treatments.

Travis Bias, DO, MPH, DTM&H, FAAFP, Chief Medical Officer, Clinician Solutions & Family Physician at 3M Health Information Systems

Like any intervention in the clinical space, there must be a focus on protecting privacy and patient confidentiality. Patients should be asked for consent prior to the usage of this technology, in line with legal requirements and organizational policies. Of course, it is fully the patient’s decision whether to employ this technology. Most patients are quite accepting once they experience the benefits.

In this new age of generative AI in the clinical documentation space, the physician’s role moves from that of creator to editor. It’s extremely important that physicians take their role as editor of the output seriously, to ensure only high-quality and accurate documentation makes it into the patient’s record. The ability for generative AI to omit key information or confidently hallucinate misinformation is well known, and we have done our best to tune our output to protect against these errors. However, the physician must ultimately vet the output of these models, ensuring the information is complete and accurate (as they do today), and sign their name to this documentation. If we deploy this technology with incredible potential only to introduce inaccuracies into clinical documentation, we will have missed a huge opportunity.

Generative AI is quite impressive in a lot of ways, so it is generating a ton of excitement. However, there is still quite a disconnect between the optimism of healthcare leaders and skepticism – or, perhaps, caution – of frontline clinicians. Both sides are coming from a reasonable place. First, I believe stronger adoption will come from leveraging generative AI for the right use cases. If used in reasonable ways, health and technology leaders can demonstrate to clinicians that generative AI will be leveraged to augment their work and to satisfy the repetitive or annoying tasks, rather than replace the rewarding pieces.

Secondly, much like other breakthroughs, I believe it will take clinicians experiencing the benefits for themselves to incorporate these new tools into their daily work. There is already a lot of excitement around ambient documentation and once a clinician sees a colleague using this capability, they will want it for themselves. Clinicians are clamoring for help with their administrative tasks, so I do not think it will be long before most clinicians are using ambient clinical documentation tools regularly.

Anish Patankar, Senior VP, GM Healthcare Software Business at Elekta

The reluctance to adopt ambient clinical voice technology stems from the classic human struggle of balancing the comfort of familiar routines with the potential for innovation and improvement that new technologies promise. Clinicians say their major barriers to adoption include: fear of change after years of established practices, apprehensions about patient privacy with the devices, doubts regarding the technology’s accuracy and the prospect of subsequent rework, fears of causing disruptions in clinic operations, and the significant challenge of allocating valuable clinic time for training amidst concerns over ongoing support.

So many good concerns to keep an eye on for ambient clinical voice technology! Huge thank you to everyone who took the time to submit a quote for us and thank you to all of you for taking the time to read this article! We could not do this without your support. What are some risks and concerns that you can see with ambient clinical voice technology? Let us know either in the comments down below or over on social media. We’d love to hear from all of you!