By
The scribe profession has existed for as long as writing has existed. From ancient Mesopotamia to ancient Egypt, scribes were high-status people documenting events and performing administrative tasks. Their mastery of literacy would lead them to specialise in fields such as geography and medicine.
To this day, this profession has persisted in the latter field. Medical scribes shadow clinicians during their consultations to document the encounter in real-time which is then reviewed and signed off by the clinician. Over time, there have been some changes to the practice, such as moving from writing on papyrus to inputting data into electronic health records. The concept of virtual scribes, where a scribe works remotely, has even taken off.
However, scribes are set for a major technological evolution in the age of artificial intelligence (AI). Recent years have seen the advent of AI scribes and they are likely to become more ubiquitous in the digital health age. To get a better outlook on AI scribes, in particular, medical AI scribes, this article dives into the concept and contemplates 3 ways in which the technology can aid clinicians and even human scribes in their tasks.
What are medical AI scribes?
Simply put, medical AI scribes are software that automate clinical documentation during clinician-patient encounters. They are algorithms that rely on speech recognition and large language models to automatically transcribe the conversation during those encounters to provide clinical notes. AI scribes perform this in real-time, by “listening in” on conversations, such as via the microphone of a smartphone or computer.
As they are based on sophisticated models, they can provide impressively accurate and streamlined notes that filter out non-essential dialogues. While AI scribes can sound similar to voice-to-text apps, they are more powerful than the latter. Developers are working to implement features beyond note-taking to their AI scribes, such as referrals and clinical decision support.
“Clinical documentation is not a goal per se; it’s the lingua franca bridging various downstream processes,” Alex LeBrun, CEO of medical AI scribe Nabla, explained in an interview. This is why the term “ambient AI scribes” has also been used to describe such tools, as they will soon be able to perform more than documenting.
However, the advent of such tools does not mean that human scribes will be made redundant. Human scribes can have a better context than AI scribes. They can inform clinicians about specific consultation details or access important information without disrupting an ongoing consultation.
In fact, human medical scribes and clinicians can benefit by collaborating with AI scribes. In the next section, we consider practical ways in which this can be the case.
Practical ways AI scribes can aid medical practice
With the recent boom in AI products, several companies have come forth with medical AI scribe solutions. In fact, investments in such products have doubled in 2024, according to the Financial Times. They are already being trialled by several institutions such as the UK’s National Health Service and Stanford Medical School.
While the adoption of medical AI scribes is relatively recent, some studies have shown their benefits. Below, we share a collection of these potential advantages that such tools can bring to medical practice.
1. Reduce the risk of physician burnout
Operating electronic medical records (EHRs) is among the major contributors to physician burnout. They can spend an average of 16 minutes per patient encounter on electronic health records. According to a former executive of NHS England, traditional note-taking for a full clinical day can take at least two hours of typing.
AI scribes have been shown to be able to reduce the risk of burnout by automating clinical note-taking for EHR entry. For example, Nuance has reported that its DAX Copilot tool has halved the clinical documentation time for physicians. Their product is in use at over 500 US healthcare groups and documents over a million consultations per month.
2. Enhance administrative tasks and workflow
Administrative burden such as clinical documentation and EHR logs not only adds to burnout but has also been shown to be inefficient. Clinicians have to go through multiple screens to find relevant information and this can reduce their productivity and workflow. AI scribes can not only reduce this incumbrance but also make such tasks more efficient.
AI scribes can not only automate clinical note-taking, but also perform additional clinical documentation. For example, in addition to generating clinical notes, AI scribes like Lyrebird and Ambiance can generate additional documents such as referrals, prescription forms and care plans. This can facilitate the work of human scribes who can review and approve the documentation, while clinicians are also supported in workflows beyond consultations.
3. Improve patient satisfaction and quality of care
By automating routine tasks such as EHR entries, AI scribes are essentially freeing up time for human scribes and physicians. In the case of MarianaAI, the company reports that its AI Medical Scribe can save over 2.5h every day.
For physicians, this enables them to dedicate more time to and focus on patients, rather than on a screen; thereby improving the quality of care. In turn, this can improve the patient journey. Studies have shown that employing medical scribes can lead to improved patient satisfaction.
What to expect from medical AI scribes in the near future?
The convenience of medical AI scribes and the boost in investments mean that we won’t have a shortage of such tools in the near future. We can further expect them to become adept at tasks beyond note-taking to provide suggestions for follow-ups and treatment plans.
However, the efficiency of such features will require deeper integration in EHR systems to be able to tailor recommendations for individual patients. In the short term, this might only be possible by some tools that collaborate with EHR providers. But this poses additional privacy risks. Stringent safeguards need to be enforced to ensure the transparency of these tools, for example regarding how they handle sensitive patient information.
Furthermore, the accuracy of these tools needs to be ensured. There are reports that suggest that about 90% of notes generated by Nuance’s AI scribe during trials at Stanford Medical School had to be manually edited for inaccuracies. Other studies have found that some AI transcriptions can even include harmful hallucinations, or made-up content.
There should, therefore, not be an overreliance on such AI output. This is where a collaboration with human AI scribes can be beneficial. With their expertise, they can review the AI-generated content, add better context, and facilitate communication between patients and medical teams.
The popularity of medical AI scribes should serve as a wake-up call for policymakers to step up their regulatory practices. The Royal Australian College of General Practitioners provides some guidelines for medical AI use but as these tools become more powerful, such guidelines need to keep up with the pace of development.
As such, in the near future, we can expect more AI scribes with additional features as well as (or, at least, hope) for adequate safeguards for their efficient implementation in practice.
Written by Dr. Bertalan Meskó & Dr. Pranavsingh Dhunnoo
