With physician burnout increasing year by year, and artificial intelligence demonstrating promise as a potential solution for taking over administrative responsibilities from physicians, there appear to be two types of solutions available.
Throughout this article, my aim is to discuss two major categories of solutions offered by artificial intelligence, as well as the advantages and disadvantages of each solution for physicians.
First, there is the solution known as AI Scribe or Ambient Scribe. In this case, physicians would record their conversations with their patients, and artificial intelligence (AI) would handle all the documentation, which means that the AI would create a summary of the conversation in the form of physician notes, or add structured information within the EMR.
The second category of solution is offered as a specialized virtual assistant. You may imagine Siri or Alexa as a medical encyclopedia, as an on-demand repository of medical terminology, and as an assistant at the point of care that performs a variety of tasks. Physicians can thus instruct it to "renew prescriptions" or "book an appointment with the patient within six weeks".
I would like to clarify before proceeding further that determining how much time each physician loses to administrative tasks and which solutions work best for them depends on a variety of factors, including, but not limited to, age, level of experience, and specialty, as well as how each physician practices - whether they prefer to document each patient's record in the room, between patient visits, or bring them home after hours.
Secondly, although artificial intelligence and natural language processing have made enormous progress, there is still a difference between their accuracy for different functionalities and skill sets. In the case of voice-to-text (which is commonly known as medical dictation or artificial intelligence dictation in the world of medical scribing), the system is very accurate and can achieve an accuracy rate of more than 90%. In contrast, other skills and functionalities like generating a summary of a conversation between a physician and a patient are not performing at a level that would suffice for medical use.
Keeping these considerations in mind, I would like to discuss the two categories of solution outlined previously:
It is not uncommon for physicians to ask us: this concept of AI Scribe or Ambient Scribe seems very interesting, but does it actually work?
There are two companies that offer Ambient Scribe in the United States; they are DeepScribe and Dragon Ambient eXperience (DAX). However, both companies offer a hybrid solution in which a human scribe and an artificial intelligence scribe work together to prepare the documentation following each patient visit. In this instance, the artificial intelligence system records the conversation and generates a summary, and sends the audio file, the summary, and the full transcription to a human scribe to validate the results and record them in the patient's electronic medical record system.
Due to this, the cost of these hybrid solutions is lower than that of a human scribe, but it remains significantly higher than the cost of other technology solutions used by physicians in primary care settings. In essence, a hybrid AI-human ambient scribe might be suitable for you depending on where you practice.
Another important thing to note about this solution is that even though they can assist with documentation, physicians are required to perform many other tasks that this solution cannot assist with. An example is retrieving information from a patient's chart. Suppose there is more than ten years' worth of medical history buried within the electronic medical record software, and doctors need to quickly access the information that is relevant to the patient's health. It would take them a significant amount of time to differentiate vital information from unimportant information in this instance.
The second category of solution offers more control and flexibility for physicians. Among the companies that currently provide the second solution are; Suki AI which is predominantly a leader in the U.S, and Tali AI leading in Canada. It is important to note that this solution does not utilize automation in the background, instead, they maintain their very nature as assistants, which means that they are available for support whenever the physician requires it.
This virtual-assistant type solution allows the physician to exercise autonomy. Several studies have been done on physician burnout, and it has been found that physicians report a lack of autonomy as a factor contributing to their burnout. Because of this, virtual assistants are an excellent solution, as they do not interfere with the practice and documentation of each physician.
Last but not least, as the category of solution is entirely AI-based, their pricing is much lower than a human scribe or a hybrid solution similar to the first category.
On a final note, it is important to remember that both of these solutions are new and developing, and it will be interesting to observe which becomes the predominant approach in the coming years as we continue to find solutions to ease physician burnout.