Katie the Intern 1/8/21

Happy second week of 2021, HIStalk! It has proven to be interesting, in the least. I hope 2021 is treating you well so far. This column is a bit shorter than usual, so enjoy an easy read! 

This time around, I wanted to expand further on the conversation I had last week with TransformativeMed’s chief clinical officer, Rodrigo Martinez, MD. We talked about the concept of AI assistants, “spot solutions” in a clinical setting that aid in the treatment plan of patients. The idea for AI-powered assistants came first from a basic question — how can we help reduce some of the frustration that exists as part of the transcription process of being a physician? 

“The first phase for a lot of these [AI assistants] is that they are listening to the conversation between a physician and a patient,” Dr. Martinez said. “The ultimate intent of it is to transcribe that conversation, but also insert the different elements of text into the appropriate spot in the medical record.” 

That isn’t exactly easy to implement, though, as a large part of this AI software is reliant on machine learning. Say a patient and provider are both speaking in a room, and the AI assistant is listening through a microphone. In the initial phases of using this software, another person would have to be present to transcribe what is happening so that the machine can learn how to do so by itself. 

“It has to start to parse out, OK, when is the doctor speaking? When is the patient speaking?” Dr. Martinez said. “The machine learning over time starts to match and map and learn using natural language processing and converting a lot of those concepts. Rather, the algorithm starts to recognize it.” 

But the transcription doesn’t stop there. The AI assistant has to then learn how to place all of this information into a medical record correctly, learning where to place orders for future visits versus when the patient and doctor are talking about past symptoms. The software must take this information in in real time before converting it into orders. 

The concept of teaching AI to listen to a conversation and decipher it is not new. AI listening and problem solving software has been researched as early as the 1960s with Newell and Simon’s General Problem Solver. For healthcare IT, the application of AI-powered scribes has come leaps and bounds. 

A leading AI offering for this application is from Saykara, known as Kara the virtual assistant. Kara was developed to help eliminate charting and billing records after hours for OrthoIndy. Saykara works by pulling patient lists to capture, interpret, and transform data from conversation and summarization to then put it into the EHR. These types of services show a reduction in provider stress levels and eliminates after-hours charting. 

I asked Dr. Martinez if AI-powered assistants have helped relieve some of COVID-19’s burden. He said the potential to reduce contact with patients and provide hands-free care is promising. When it comes to telehealth and virtual appointments, AI assistants and scribes could grow to be the future of healthcare. 

“As more and more visits are moving to telehealth, there is a role for that AI-powered scribe to be recording that information and applying it,” Dr. Martinez sad. “I could see how the more and more you make things hands free, and make it easier to anticipate the next step that the nurse or physician needs to take, the better off folks are going to be.” 

The AI-listening concept may be off to a slow start, Dr. Martinez said, but the potential for outpatient treatment is more promising than in-person. In-person treatment usually has too many voices and people in a room for scribes to accurately transcribe. Ultimately, the future of these AI assistants lies in the hands of further development and machine learning.

Dr. Martinez said that the future of AI in medicine is promising, and that AI can be applied to a multitude of areas in healthcare IT and telemedicine. AI has the potential to make transcription and hands on care easier and safer, taking a weight off of healthcare providers in general.

“I think the concept of that is very fascinating and very interesting,” Dr. Martinez said. “I think there are a lot of potential applications to decreasing the administrative burden that is part of healthcare in the US.”

That’s it for this column! I hope you’re enjoying 2021 to the best of your ability, and stay safe out there! 

Katie The Intern

Email me or connect with me on Twitter.