Disclosure: Benjamin H. Levy III, MD, has disclosed no relevant financial relationships.
Artificial intelligence (AI) technologies are emerging to help physicians, nurse practitioners, and physician assistants write notes in real time using dictation software. This is extremely important because note writing for electronic medical records (EMRs), or electronic health records, is a major contributor to physician burnout.
After obtaining consent from the patient, electronically in the EMR, in a paper form, or after the patient gives verbal consent in the clinic room, the provider can turn on the digital scribe application, sometimes on their smartphone. It will listen to the clinic visit and write a Health Insurance Portability and Accountability Act (HIPAA)-secure, cloud-based note.
AI knows how to separate out information provided by the patient to go into the history of present illness (HPI).
AI technology will listen for medications, social history, physical exam, and assessment and plan, and put this information in the correct place and format as well.
At the end of each patient visit, physicians can still edit the note, but the AI software dramatically reduces the amount of time spent typing, so that physicians can focus on the patient conversation and talk in a much more natural way. Physicians will no longer have to ask patients to slow down their description.
Intelligent documentation support software uses speech recognition, natural language processing, and AI to write notes and to help automate tedious clinical documentation, which physicians and medical providers spend hours each day typing out.
This AI software, produced by several different companies, now allows physicians to focus their attention on data gathering, asking important questions, and trying to develop an assessment and plan instead of being bogged down by typing.
Administrative tasks such as typing in the EMR and paperwork are one of the biggest causes of physician burnout.
A 2016 study published in the Annals of Internal Medicine found that physicians allocate approximately 49% of their time on EMR and desk work. So, intelligent documentation has the potential to make clinics more efficient and allow physicians to see more patients each day.
A very interesting 2023 study looked at 307 primary care physicians who practiced medicine across 31 primary care offices at Massachusetts General Hospital and Brigham and Women's Hospital in Boston, Massachusetts. It was reported that for every 30-minute patient clinic visit, physicians spent 36.2 minutes on the EMR; 6.2 minutes of nighttime work from home; and 7.8 minutes on their electronic inbox to follow-up on labs, pathology, and patient questions.
Medscape's 2022 Physician Burnout & Depression Report found that 60% of physicians reported that having too many bureaucratic tasks (eg, charting, paperwork) contributed to feelings of burnout.
AI technology to assist with EMR documentation has the potential to decrease physician burnout and to provide highly accurate notes that include patient descriptions of their HPI.
The AI software remembers dates and numbers easily.
This is helpful in a gastroenterology clinic when patients are describing, for example, how many bowel movements per day they've been experiencing if they have constipation or diarrhea, or specific dates they experience rectal bleeding. Another example might be a particular date for an emergency department visit that the patient mentioned very quickly in passing.
When using the dictation software, I try to summarize CT scans, MRIs, and ultrasounds. Before each impression, I'll mention the date of the imaging, and the AI software will organize all this information into the note. After the AI technology develops a well-organized note, you can still edit by directly typing any aspect of the note you'd like before citing.
Software such as Abridge, DeepScribe, ATLAS.ti, Ambience Healthcare, Nabla, Suki Assistant, Nuance Dragon Ambient eXperience, Copilot, Oracle Health Clinical Digital Assistant, and Freed have the potential to improve the delivery of healthcare by allowing physicians as well as nurse practitioners and physician assistants to focus on patient conversations. Many other AI dictation systems have recently launched, which will hopefully help keep driving the technology forward.
Whenever possible, I recommend that physicians request a pilot trial of the software before expanding system wide.
There are many pros for AI-enhanced dictation software, including helping providers finish their notes during clinic hours and reducing the time spent outside of work completing notes.
AI can speed up the process of helping physicians write after-visit summaries too, which are important for keeping patients organized.
For providers who haven't taken a keyboarding class or mastered hand placement on the keyboard, AI dictation can significantly improve the ease of writing notes.
Physician and public health leaders hope that by reducing the requirements for typing, AI clinical documentation technology will prevent musculoskeletal repetitive stress injuries such as carpal tunnel syndrome and de Quervain tenosynovitis, an inflammatory condition causing pain and swelling in the tendons between the wrists and thumb. Our hands will feel less tired at the end of the day.
Some patients have told me that they like AI documentation because it illustrates that a healthcare organization or private practice group is invested in modern technology that improves clinical care.
There are some potential negatives or reasons that certain physicians might hesitate to adopt AI dictation technology.
This can be true for gastroenterology subspecialists; for example, those focused on managing inflammatory bowel disease. For these subspecialists, the AI software might not include every single detail, from documentation of the patient's extensive colonoscopy to their medication history, in exactly the same way that they prefer in their documentation. Sometimes the AI can make histories a bit too generic.
However, over time, I fully expect that AI-assisted electronic dictation software companies will fine-tune this technology to include the extended content frequently written by subspecialists. It's very important for physicians to work with their information technology departments and the software company prior to launch of this electronic documentation software to fine-tune specific preferences.
Another potential negative for using this software is the time that it takes to dictate long paragraphs from CT or pathology reports. In these circumstances, I recommend that physicians consider copying and pasting the most important findings from these reports into the notes separately. This can easily be done by using AI dictation for the rest of your note.
From my experiences, the AI software works well for common gastrointestinal problems, such as constipation, heartburn, abdominal pain, and diarrhea.
Sometimes the AI software might not include a small section of your dictated plan. Therefore, it's important to read through the entire dictation and make any necessary edits. During my notes, I try to give the AI lots of directions by saying, "include this in the plan," which definitely seems to help.
A 2021 study by Wang and colleagues reported that, "the digital scribe was about 2.7 times faster than both typing and dictation. For physical exam sections, the digital scribe was about 2.17 times faster than typing, and about 3.12 times faster than dictation."
In addition, digital scribe software is much more efficient, transcribing the note in 10-20 seconds instead of waiting hours. Literally, you click "create note," wait a couple seconds, and it's done.
A major goal for digital scribe software is to give healthcare providers the ability to dictate in a manner less expensive than traditional transcription services, allowing them to save money.
In medicine though, we have to be cautious about choosing the right kind of AI-driven digital scribe application for clinical documentation. Not all AI is created equally. The software should be created specifically for medical use.
For instance, a recent study by Baker and colleagues in the Journal of the American Academy of Orthopaedic Surgeons showed that ChatGPT could "improve clinical documentation by producing more comprehensive and organized HPIs." However, the double-blind study also showed that "ChatGPT included erroneous information in 36% of the documents."
So, it's definitely a good idea to stick with validated software programs specifically designed for medical care. In its current form, ChatGPT would not be an ideal choice.
It's also a good idea for healthcare organizations to select AI software that will delete the audio recording at the end of the patient visit. From personal experience, I would say this is the most frequent concern among patients who have started using this AI-driven software during clinic visits.
It's very important to explain the AI dictation security at clinic check in and again during the beginning of each patient visit. It's also a good idea to include a sentence in the clinic note referencing that you have permission from the patient to use the digital scribe software.
Amazingly, some future AI applications in development include the ability to automatically generate billing codes from intelligent dictation documentation software. This billing technology is starting to emerge using a mixture of machine learning and natural language processing.
Over time, I expect AI to help physicians and medical providers become more efficient, with less clicking required in the clinic. I also expect surgeons to take advantage of this technology while performing operations, such as the ability to rely on real-time dictations during endoscopy procedures.
AI-powered clinical documentation software has the potential to revolutionize the way physicians, nurses, and physician assistants take care of patients. From personal experience, this AI software allows physicians the opportunity to focus on patient care, without the physical and mental energy drain of typing notes all day long.
This technology also may help clinicians build rapport with patients more easily, allowing us to look patients in the eye, have more natural and free-flowing conversations, and express empathy more easily instead of having our heads buried in the computer typing away during a clinic visit.
