DeepScribe

So today we’re chatting about DeepScribe and how it’s used at your company. Before we jump into that, could you give a brief overview of the company and your role there?

I’m a vice president and CIO of a health system. We have an independent hospital with around 5000 employees as well as a few hundred physician offices to support the community. I’ve been in this role for over eight years.

What was the need that drove you to look for a solution like DeepScribe?

The “why” behind this initiative was to address physician burnout by reducing the burden of documentation. We analyzed our physicians’ usage of the Epic system and found that they were spending the most time on notes and documenting. So we focused our efforts on improving physician documentation by exploring various tools and technologies, including virtual scribes and ambient AI. About two years ago, we started piloting different vendors’ products, and a little over a year ago, we selected and implemented DeepScribe as a pilot with 10 providers. Now, we have expanded it to about 28 providers.

What were the key requirements you were using to evaluate vendors?

The three primary requirements were a quick turnaround time for the note, high-quality notes that require minimal editing, and seamless integration with the Epic system. In terms of integration, the basic level included visibility into the patient schedule and automatic appearance of the AI-generated SOAP note in Epic. The second-level integration allowed the use of existing templates within Epic while incorporating components of the AI-generated note. This ensured that providers could leverage the templates they had already created while still benefiting from AI. The AI can help better capture the uniqueness and complexity of a patient visit and add that to the existing templates.

Are those templates on the provider level?

The doctors use templates that they’ve created. During the training process, which takes a couple of hours, they learn how to incorporate the AI-generated paragraph into the template. In the EHR system we use, Epic, this is called a Smart Text Link. DeepScribe offers several of these Smart Text Links that can be inserted into existing templates. When the conversation between the doctor and patient is transcribed into an AI-generated note, the relevant text is automatically inserted into the appropriate section of the template using the Smart Text Link.

What other vendors were you comparing against DeepScribe?

The two main competitors we looked at were Nuance DAX and 3M M*Modal’s Fluency Align.

How did they compare to each other, and what made you decide to go with DeepScribe?

DeepScribe stood out among the other options we considered for a few key reasons. First, the turnaround time for the notes was crucial for us, and DeepScribe was able to deliver them within a couple of hours consistently, making them a leader in that aspect. Second, the quality of the notes was of utmost importance, and after our providers trialed it, they felt that DeepScribe provided the highest quality notes with a quick turnaround. And we’re a bit biased. as we were already a development partner with DeepScribe, but they had great integration with Epic. We worked closely with DeepScribe to develop the interface they offer customers now.

How did the pricing models compare across the different options?

Each vendor has their own pricing structure. We found the tiered model based on volume with an unlimited option to be particularly appealing for us. This flexibility was unique across the different vendors. Nuance and 3M M*Modal both offered a cost per encounter pricing model, so they provided an estimated cost rather than a definitive rate.

How would you characterize the sales process with DeepScribe?

We had a close working relationship with them during the integration process, which made it a different experience than others might get. It felt like we were partners in developing the product together, which was really nice. Being one of the first to integrate with Epic gave us direct access to the vendor’s executives and the ability to provide feedback. They were great to work with and overall a good company.

Did you do a pilot with them before deciding to move forward?

We conducted a pilot around three years ago but found that the technology wasn’t quite ready at that time. About a year and a half ago, we revisited the idea with DeepScribe and found that their AI technology had progressed significantly. Our providers were impressed with the maturity of the AI, so we piloted it again, but without integration. Our main focus during the pilot was on turnaround time and note quality, so integration wasn’t necessary to assess those aspects. The providers copied and pasted the note from DeepScribe’s portal into Epic after it was created. Once we finished the pilot phase and were satisfied with the results, we started working on the integration. By Q3 of last year, we were fully live with integration, out of the pilot phase, and had signed a contract for the product.

How was the onboarding and setup process?

DeepScribe heavily engaged with us throughout the entire process, also collaborating closely with Epic. We had numerous calls with Epic, and DeepScribe also made multiple on-site visits to ensure a smooth onboarding process for our initial providers. Their support included addressing any optimization requests and ensuring that the system was functioning properly. Currently, we have over 28 providers who are live on the system. Thanks to DeepScribe, onboarding new providers is now a straightforward process, which includes a couple-hour session of education. They have also provided us with the capability to add users and set up basic functionalities.

How do you use DeepScribe in your workflow?

We give our providers Wi-Fi-only iPhones with the DeepScribe app for recording patient visits. Before entering the patient room, they do a quick recording that serves as a pre-visit dictation. They then inform the patient that the visit will be recorded and obtain consent. During the encounter, the iPhone records the patient/provider interaction. After the visit, some providers do a third recording to summarize their perspectives. The recordings are then sent to DeepScribe for transcription. The AI algorithm converts the transcript into a word-for-word record, which is then turned into a SOAP note. The notes are reviewed by humans who work for DeepScribe, and they’re then pushed into the Epic system, where providers can review and sign them, which completes the note, and it then goes through billing using the traditional process.

The full transcripts and recordings are stored in the DeepScribe portal for 30 days, but the legal record is the one in Epic. And they’ve recently started offering fully AI-generated notes, which have a quicker turnaround time. The quality is good, and we’re gradually moving in that direction. Our focus was initially on primary care, as we thought it would be the most challenging undertaking, but DeepScribe now supports all subspecialties, and we’re planning to expand beyond just primary care.

What do you see as a weakness with DeepScribe?

I’ve talked to DeepScribe about evolving their product to assist with coding to help maximize reimbursement for healthcare organizations. This would help justify the cost of the product and provide a key return on investment. While the current focus is on easing physician documentation burden, there is a lot of financial pressure in healthcare, and having help with coding would be a valuable addition to the product.

Has DeepScribe achieved what you hoped it would?

We use Epic as our electronic medical record. One great feature is the visibility it provides into doctors’ note-taking activities. We can track how long they spend on notes and how much time they spend in the system outside of regular office hours. Our data shows a significant reduction in the time spent on notes. Most of our providers are saving 20-30 minutes per day in documentation, and we have the data to prove it.

We also prioritize the well-being of our doctors. We surveyed them, and the feedback has been overwhelmingly positive. Their qualitative indications show that they find it beneficial. In addition to the subjective feedback, we have quantitative data that backs it up, showing a reduction in time spent on notes and outside regular business hours, so we believe it’s doing what we hoped it would.

What made you decide to go with an external product rather than waiting for Epic to develop this functionality?

Epic made it clear that they were not going to develop their own ambient AI product. Instead, they planned to partner with third-party products and integrate them into their system. DeepScribe was one of the top vendors that Epic was working with for deep integration, which is why we were evaluating them, and we selected DeepScribe as our preferred vendor because they had the best performance in terms of turnaround time and note quality.

What do you see as an area of growth for DeepScribe?

The company needs to work toward developing a fully AI product without human intervention. There is a competitor, Abridge, that is slightly ahead in this area. Currently, we have three providers on the fully AI product, and testing is progressing well. However, there are a few missing functionalities that they are actively working on. One important feature that many providers like is problem-oriented charting, where the problems are listed and documentation is provided under each problem. This functionality currently exists in the human-in-the-loop product but not in the fully AI product. They need to prioritize adding this feature and ensuring its effectiveness. It’s important for them to continue developing the product and establish themselves as a leader in the dynamic healthcare vendor market.

Have you found the platform to largely be reliable and stable?

We’ve had very few issues. We track turnaround times regularly, and we occasionally experience a delay in the human-in-the-loop QA process, resulting in notes taking around five hours to be completed. When this happens, we notify DeepScribe and they quickly resolve the issue. They also have a support feature within the app where we can chat with a representative in real-time. If there is an issue, we simply inform them through the chat, letting them know that we are waiting for a note to be completed. Overall, the reliability is quite high, with only a few notes needing additional escalation.

How would you describe the support that you’ve received from DeepScribe?

We’ve had a close working relationship with them, and they’ve provided us with additional support whenever we’ve encountered any issues. They also offer in-app support, allowing our providers to chat with a live support agent if any problems arise. Overall, their support has been very good.

Do you feel you made the right decision in choosing DeepScribe?

Yes. Conversational ambient AI technology has become increasingly prevalent and has quickly become a valuable tool. It’s been impressive to see how much it has permeated the industry in such a short time. Because of that, we’ll need to continuously evaluate whether DeepScribe remains the right decision for us going forward. But they’ve been excellent to work with, and I would definitely choose them again if given the chance.

Do you have any general advice for folks who are going through this decision-making process?

When considering implementing a solution like DeepScribe, it’s important to think about the financial implications. This added expense needs to be justified, especially for highly productive physicians. You’ll need to determine how to make the financials work and whether increased patient volume is needed to justify the expense for less productive physicians. One approach is to highlight that using DeepScribe is much less expensive than hiring a human scribe. However, it’s crucial to understand that, even though DeepScribe is cost effective compared to other solutions, it still represents a significant expense for healthcare providers, especially given the current healthcare environment. You’ll need to take into account the difficult economics and the fact that most healthcare organizations are struggling to generate consistent margins. So if you’re positioning DeepScribe as a provider wellness initiative, you’ll have to consider the financial aspects as well.