Ambient AI scribe technology: The future for clinical documentation?

© ipopba - stock.adobe.com

Electronic medical records were supposed to make charting easier. In some ways they have, however, medical charting is still a time-consuming chore for medical providers who would rather be spending their time with patients or family. Enter artificial intelligence (AI), and the potential for safely integrating AI functions into medical practices to improve workflow.

A recent study on the viability of using ambient AI scribes for clinical documentation, published in the New England Journal of Medicine, indicates a sea change for physicians and practices seeking to utilize the evolving technology to alleviate charting burdens. The bottom line: AI scribing is accurate, easy to use and improves patient interactions.

Addressing the challenges of clinical documentation

According to the study, increasing charting demands have necessitated more "pajama time" for clinicians. While this may sound comfy, it actually means healthcare providers are spending substantially more time on clerical activities outside working hours. This extra administrative workload is creating unprecedented levels of burnout that are threatening the stability of the physician workforce.

Assessing the effectiveness of ambient AI scribe technology

To tackle this urgent challenge, a multidisciplinary physician group with over 9,000 physicians located in Northern California did a rapid regional pilot assessment utilizing ambient AI scribe technology. Their goal was to identify optimal approaches for using ambient AI scribes to safely, and effectively, improve workflow.

The researchers used an ambient AI scribe platform that utilizes machine learning to transcribe clinician-patient encounters in real-time and instantly convert speech into text. The AI scribe software then applies natural language processing techniques to summarize key clinical information and produce SOAP notes, which significantly reduces time spent entering and editing clinical documentation.

Basically, the pilot set out to do the following:

• Demonstrate the viability of growing and sustaining adoption of ambient AI scribes, based on the number of clinicians and the percentage of patient across diverse specialties and settings
• Maintain documentation quality by assessing the best approach for safely integrating ambient AI scribing into practice protocols

Thorough training and implementation key to success

The pilot program leveraged the practice’s extensive experience in large-scale technology integration to train the users and provide support as they got used to new processes.

The physicians assessed the quality of the AI-generated documentation using a modified version of the Physician Documentation Quality Instrument (PDQI-9), with transcripts averaging a score of 48 out of 50.

Results indicate strong potential

Ten weeks post implementation, 3,442 physicians had used the ambient AI scribe in 303,266 patient encounters across various specialties and locations. Physician response has been overwhelmingly positive. Providers appreciate being freed up to have more personal, effective patient interactions. One physician remarked, "It makes the visit so much more enjoyable, because now you can talk more with the patient and concentrate on their concerns."

Patient feedback has also been favorable. A sample of patient surveys indicated 71% of patients felt they spent more time speaking with their physician, and 81% reported their physician spent less time looking at the computer screen. All patients surveyed felt comfortable with AI tools being used during their visits.

Key takeaways

The multidisciplinary pilot shows key early indicators for success:

• Clinicians noted AI scribes have the potential to significantly reduce the time spent constructing clinical notes.
• Ambient AI scribes were found to be acceptable for clinicians and patients, largely improving the experience for both parties.
• While AI scribes can produce highly accurate transcripts, they may still have inconsistencies that require physician review and editing.

Given the incredible pace of change, building a dynamic evaluation framework will be essential for assessing the performance of AI scribes across several domains including engagement, effectiveness, quality, and safety.

Terry Ciesla is senior vice president of ScribeEMR in Woburn Massachusetts