Talking HealthTech: 328 – Speech Recognition Technology in Healthcare – Simon Wallace, Nuance

Meet Dr Simon Wallace

Dr Simon Wallace leads the Healthcare division at Nuance in the United Kingdom and Ireland as the Chief Clinical Information Officer (CCIO).

About Nuance

Nuance is a global organisation that provides cloud-based speech recognition solutions powered by AI to assist clinicians with clinical documentation. Nuance was recently acquired by Microsoft, which is a really exciting next step in its speech recognition journey. 

Nuance is made up of two parts—one of them is the enterprise side, which deals with omnichannel conversations and using voice biometrics when you authenticate yourself to a bank or to the tax office in the United Kingdom. On the healthcare front, Nuance has Dragon Medical One.

Dragon Medical One, Nuance’s speech recognition solution, is being used by over 550,000 clinicians in a variety of countries and languages.

The Clinical Documentation and Admin Burden

Throughout the last couple of years, one of the big themes emerging, even pre-pandemic, is the stress and strain healthcare professionals are under. The pressures of waiting lists, hospital capacities, and the like; and of course, COVID came along in March 2020 when the UK was locked down, and the health service did very well and provided an excellent service. Still, of course, after a number of shutdowns due to COVID waves, there is a massive backlog of patients for routine hip-knee operations, but also more serious treatments for cancer and heart disease. These healthcare professionals have yet to really have a break.

Clinician stress and burnout can be partially attributed to the time spent on paperwork and administration. Professor Cyril Chantler coined an excellent phrase: “Medicine used to be simple, ineffective, and relatively safe. It is now complex, effective, and potentially dangerous.” Safe medical care requires accurate and complete clinical records. In the case of COVID, no one could have predicted that a lack of taste and smell would be symptoms or that dexamethasone potency would be an absolute game changer. Still, it required quality documentation to understand that those were really important factors. 

Nearly a thousand clinicians were surveyed by Nuance to determine how much time is spent on clinical documentation; they found that, on average, 13 and a half hours of a 37-and-a-half-hour clinical week are devoted to the documentation process alone. According to a comparison between this and a survey conducted in 2015 by Nuance, this figure is 25% higher. Another striking finding was that the average clinician spends about 3.2 hours per week, and the average consultant spends about 4.5 hours per week outside of work on documentation. To sum up, clinical documentation is crucial, but it places a heavy burden on day-to-day operations and prevents doctors from focusing on the most essential part of their job: interacting with patients.

Benefits of Dragon Medical One

From a speech recognition standpoint, there are a number of things that really make it an excellent digital tool for clinicians to use, but the most important thing is that it will solve the problem that the clinician is facing.

Talking is three times faster than typing. As a result, you will have more time on your hands. The accuracy is what matters the most. Ten or twelve years ago, some people might have used speech recognition, and there might have been some correction of odd words and all that sort of thing. In the years since, advances in AI, deep learning, and algorithms have allowed speech recognition software like Dragon Medical One to achieve truly remarkable levels of accuracy when used in tandem with a comprehensive medical dictionary. In recent years, that has been the game changer.

The fact that Dragon Medical One is hosted in the cloud is significant for a number of reasons; however, the clinician will appreciate, first and foremost, the fact that speech recognition can be used immediately after installation. The days of training speech recognition by reading it a few paragraphs and having it put to use are long gone. It’s ready to go right out of the box and into your hands. It’s a never-ending cycle of profile refining, document revision, and learning from previous iterations. That leaves you with a personal profile that is unlike anyone else’s. Another perk of working in the cloud is that you can use any device, not just a desktop or laptop. It’s possible to transition from an operating room to a clinic, an office, or another hospital, and many medical professionals did so during the pandemic. Therefore, the cloud-based feature provided a great deal of operational leeway.

Because typing takes time, you’re more likely to take shortcuts and leave out important information that should be included in your notes, but the result is often a more thorough and complete document. To cite just one example, consider the use of pertinent negatives. Usually, only the positive things would get typed up, but it’s crucial to record even the absence of negative symptoms, such as a cough or wheeze. However, by using voice, you have more time to do so. It’s perfect because of the completeness of the note, but if there was some medical or legal action and some negative wasn’t recorded, no one knows that you actually asked that question, and it’s not there. As a result, the quality of the note improves.

Voice can also assist in navigating an electronic patient record (EPR). You can use voice commands instead of clicking and tabs to perform a variety of functions. A few months ago, Nuance developed a voice command for a urologist to perform a routine renal ultrasound report on 75% of his patients following an outpatient clinic visit. Nuance created this step-by-step command that walks the commission through the various tabs and clicks he would have to do manually in the electronic medical record (EMR). The urologist gave a single voice command: request a six-month renal ultrasound, and everything happened right in front of him. He went on to say that he saved himself about a minute and a half compared to doing it the traditional way. For ten patients, that’s an extra 15 minutes each day.

Cultural Shift

At present, we are going through a cultural shift. When computers came along, we began to document our notes electronically, and now we’re shifting from the keyboard to voice. We use our voice in many non-work, everyday functions, and many health professionals wonder, “Why can’t I use that at work?”

The use of speech is beneficial in primary care because it allows doctors more time to deliberate on their clinical decisions. Furthermore, patients are frequently present in the room and appreciate hearing about the processes that are taking place. According to research, patients only absorb about a quarter or a third of what is said, and some of the feedback that doctors have received indicates that patients value having more exposure to the conversation that was had.

The Role of AI in Speech Recognition

Speech recognition is a foundation upon which other AI, cloud-based speech recognition solutions can be built. For example, one of the EMR’s frustrations is the clicks and moving around. The EMR is very important because it aids in clinical decision-making, support, analysis, and timely document presentation, but it does take time. Where technology is now going, an individual’s voice is used to navigate much more around the EPR.

Ambient clinical intelligence is when the desktop in the clinic room disappears, and a smart device on the wall with numerous microphones and sensors is used to capture the conversation between the patient and the doctor. So, voice biometrics will begin by allowing the doctor to sign in using his voice because each of our voices has a unique identifier. And then voice biometrics dire raises the conversation. Everything that is spoken is documented. Then, at Nuance, they have a smart natural language processing called clinical language understanding, which picks out the key clinical terms, key medicines, and key components of that conversation. It converts it into a codified structured note so that when the clinician is finished with the patient, he can use a voice command to bring up the note for which he would’ve used his voice. But it was done in the background using natural language processing. Then check it for accuracy to ensure that it is what they want to send out, and then simply issue a voice command.

The Future of Nuance

Moving forward, the shift in cultural norms will be the most important thing for Nuance. Within the next one to two years, it will make it its mission and spearhead its crusade to convert as many healthcare professionals as possible to speech. When we talk about healthcare professionals, this encompasses not only physicians and nurses but also all of the allied health professionals. This is due to the fact that having that cultural shift for a move from the keyboard to voice then provides that foundation for the fascinating AI stuff. 

_{Source talkinghealthtech.com}