DeepScribe’s AI-Powered Solution to Physician Burnout Could Also Help Eliminate Medical Errors

January 11, 2022
8 min read
Featured Image

Even before the pandemic, physician burnout* was a prevalent problem in health care – due in no small part to the immense number of hours doctors have to spend maintaining patient health records. This daily administration negatively impacts job satisfaction among both residents and attending physicians because it greatly reduces time for patient care. It can also increase medical errors.

As the son of a physician, Akilesh Bapu, who co-founded DeepScribe along with UC Berkeley classmate Matthew Ko, has first-hand experience with the challenge this presents. During his teenage years, Bapu watched his father spend most of his free time on nights and weekends transcribing notes from patient exams he conducted during the workday. That intense focus and commitment to detail unquestionably reduced errors that could occur during the transcription process, but also eroded his time with his family – and his love for medicine. 

The younger Bapu could see that burnout was a constant threat for his father as his administrative burden increased over time. So, as a PhD candidate studying natural-language processing at Berkeley’s AI Research Lab, he started to look at how AI technology could make a difference for people like his dad . 

Dictation software was one option he considered, but it wasn’t accurate enough for life-and-death subject matter. It was also prohibitively expensive. AI scribes required specific verbal prompts from users during dialogue that could derail the communication between doctors and patients. They are also error-prone, missing linguistic complexity and sometimes failing to capture both sides of a conversation. 

Beyond technology, human scribes are also an option, but they take months to train and are just as prohibitively expensive. 

What Bapu and Ko wanted to create was an affordable, AI-powered solution with accuracy that surpasses human scribes – a “scribe for the masses,” as Bapu puts it. 

The resulting solution has grown DeepScribe from a team of three employees serving ten clients when they launched in March of 2020 and 35 employees serving 500 clients today.  Bee Partners led the company’s $5.5M seed round in May 2021 and DeepScribe has just announced a Series A round of $30M.

The proprietary technology that puts the “scribe” in DeepScribe is only the first step in a commitment to improving the lives of millions of patients … and their providers. In the process of solving the problem for his father, Bapu and the DeepScribe team also established a way of collecting a rich and unique data set that physicians can use to improve their diagnoses and avoid medical errors. DeepScribe now securely houses millions of queryable data points that can yield previously unavailable information for more accurate diagnoses and better care. 

I caught up with Bapu recently to find out more about how the company is enabling doctors to focus on providing better care to patients – and themselves.                                                                                                                                                                                                                                     DeepScribe has a very personal origin story. Can you share that?
The burden of transcription is a problem that my dad, who is a physician, personally faced. Medical transcription was consuming almost half of his personal life on the evenings and weekends. That is what motivated me. It also affected a lot of the design decisions early on. Companies who have this kind of technology in healthcare like to see how much they can charge for it. With DeepScribe, our idea was to build a scribe for the common doctor, the 90 percent or more that are not going to throw money away on scribes and dictation tools that are not working. We had the vision of building a scribe for the masses. 

Thirty-four percent of all data breaches in 2020 were in the healthcare industry. This is obviously a big concern.  How do you ensure privacy?
We have always started with the idea of security first. Even when we only had ten or twenty customers, we went above and beyond what was required. The security standard for patient health information is encryption-in-store and encryption-in-transit. But everything we have, from patient information to meta-data that can’t be identified, is all encrypted in storage and in transit. We follow all the protocols and standards for HIPAA, and these days the providers such as Amazon have their own HIPAA programs, so that makes it seamless to follow. That being said, we hear that a lot of the data breaches happen because of the way the database accounts are set up, not because of cloud providers like AWS. From a data breach standpoint, we can confidently say we have our ground covered. In terms of the future, I think the biggest risk for us is to assure these health systems that we are using the data for the right reasons: to improve the model.

Speaking of Amazon, don’t they and Google already have technology like this? What precludes them from building a similar product? 
Amazon and Google have had a long time to build this if they wanted to; they probably encountered the problem at least a decade before DeepScribe. What they’re doing is building solutions that other companies or developers can use to solve this problem. Google is building speech text, which is one of the many components of a system like this. Amazon has gone one step further and is doing medical-entity recognition which helps a little bit – you can use that to understand the transcript. But if you put any of their solutions in front of a doctor, the doctor will tell you they don’t really solve the problem: they don’t save providers time on documentation and eliminate physician burnout. These companies are doing what they do best, but we don't see them building a full-fledged solution that is able to be purchased by a doctor or a health organization.

How do you see your opportunity in terms of company growth?
There are about a million MDs in the United States. On top of that, you have another half-a-million to a million nurse practitioners and physician assistants who also practice medicine. So if you look at how much DeepScribe costs right now for each physician and how many physicians there are in the U.S. alone, that’s enough for us to become a billion dollar company. That’s our current focus – scaling to work with every physician in the U.S. And our work is not just about transcribing technology. There’s a lot more to do with downstream and upstream tasks as well.

Beyond time efficiency, in ten years, what does DeepScribe mean to the healthcare community? 
We’ve been spending a lot of time on that question recently. Right now, DeepScribe stands for the idea that you don’t have to hire a human scribe to get rid of burnout. And therefore, for the first time, there is a scalable solution to systematic documentation issues and burnout. 

On the back end, what is actually happening is that physicians are trusting DeepScribe to listen to the most crucial point in the healthcare workflow: the physician-patient conversation. They are trusting us to enter the exam room. When that happens, we have access to the entire conversation and to the electronic health record, which means we have access to previous information. 

So, for the first time, providers have given access to a company to house patient data so it's usable in various ways. Physician burnout is the second biggest problem in healthcare. The first is medical errors. That’s what we have always planned to address. When providers are diagnosing, they can rely on their experience,  colleagues, online resources, guidelines. But they don’t have data from around the country to help evaluate how good their diagnosis is. You can’t search across electronic health records for “forty-five year-old female with experience in marathons currently suffering symptoms of heart failure,” but you can use that query on our data. We can surface insights based on data we have that no one else has.

Watch my in-depth conversation with Akilesh Bapu, founder of DeepScribe.

DeepScribe is hiring!

To learn more about DeepScribe, visit their website to sign up for a demo, or follow them on LinkedIn for regular company updates.

About Bee Partners
Founded in 2009, Bee Partners is a pre-Seed venture capital firm that partners with revolutionary Founders working at the forefront of human-machine convergence across technologies that include robotics, AI, voice, i4.0, and synthetic biology. The firm leverages a singular approach to detecting new and emerging patterns of business as well as inside access to fertile but often overlooked entrepreneurial ecosystems to identify early opportunity in large, untapped markets. Bee’s portfolio companies consistently realize growth at levels that outstrip industry averages and secure follow-on capital from the world's top VCs.

Follow Bee Partners on LinkedIn for regular updates.


No Comments.