Physician burnout is one of the most serious conditions in today’s medical profession. The Agency for Healthcare Research and Quality defines the condition as “a long-term stress reaction caused by emotional exhaustion [and] depersonalization,” among other factors. According to the American Medical Association, physicians suffer from considerable stress caused by facets of their job that have little to do with actually providing personalized patient care. The AMA reports that physicians spend up to six hours daily working with electronic health records (EHRs) to adhere to government and hospital documentation requirements. That’s six hours not spent seeing patients, and thus not having the time to listen carefully and diagnose, empathize, hold a hand, speak with family members, or explain conditions and next steps. Physicians are not to blame.
All of this belies the fact that good patient care is the primary source of job satisfaction for 79% of physicians, according to The Physician Foundation. Healing patients is why doctors choose to practice medicine. Dissatisfaction is sounding alarms as physicians leave the field at alarming rates.
At the most recent Health Information and Management Systems Society’s (HIMSS) annual conference, technology was front and center of industry efforts to improve health care. Nuance Communications’ focus on how Conversational AI and Ambient Clinical Intelligence (defined below) are particularly promising. I had the opportunity to speak at length with Nuance CEO Mark Benjamin about how technology may help cure physician burnout. The transcript of our interview follows.
How do you define Conversational AI?
Mark: That’s an important place to start our discussion. Conversational AI uses intelligence to convert speech not just into text, but into text with meaning. It understands the speaker’s intent in context, responds with relevant and helpful information, and extracts key structured data that can drive actions and analytics. Conversational AI is a significant advancement of the AI-powered speech recognition systems used every day by doctors, lawyers, automotive manufacturers, financial institutions, and consumers.
Why does Conversational AI matter in healthcare?
Mark: Conversational AI matters on multiple levels across the healthcare system. Medicine is a complex technical, economic, and regulated environment. Adding to the complexity is an aging population; not enough resources to care for everyone at scale; and healthcare insurance that is either unaffordable or inaccessible for far too many people. So, in addition to focusing on their core mission of providing high-quality patient care, healthcare providers are strategically focused on operational performance and financial optimization, and they also deal with the ever-increasing administrative burdens placed on them. That’s had the effect of placing pressure on clinicians to handle increasing workloads, leading to what the president of the World Medical Association has called a “pandemic of physician burnout.”
Doctors are doing their best with electronic medical record technology to manage a staggering amount of documentation required for insurance coverage, as well as reimbursement, medical, and legal liability, all while continuing to prioritize patient care and quality reporting. Unfortunately, doctors are now spending about twice as much time entering or re-entering administrative data into computers than they do interacting with their patients. This work is often done at night at home during what they call “pajama time” – essentially family and personal time. I think any professional can relate, but what’s different is that we’re seeing skyrocketing burnout rates and elevated suicide rates among physicians. In fact, 51 percent of physicians reported experiencing frequent or constant feelings of burnout. The situation isn’t sustainable, and people are leaving the field at precisely the time we need them most.
This not just an economic or systemic problem. Physicians practice medicine for the higher purpose of caring for people. Yet they feel they’re held hostage to regulatory and technology demands that keep them from that. Our journey began there, with the goal of getting technology out of the way by leveraging Conversational AI to help clinicians do their jobs better and faster. The broader goal is to improve physician productivity and job satisfaction, their well-being and, most importantly, clinical outcomes for patients. Our mission is all about taking care of clinicians so that they can take care of their patients.
Conversational AI can radically transform the doctor-patient relationship in the exam room. We have a saying: don’t treat the chart, treat the patient. It sounds so simple, but the power of human touch and looking someone in the eye cannot be overstated. We have data that shows there is a definitive, positive impact when doctors make eye contact or hold an anxious patient’s hand. That connection is lost when doctors must turn away to face a computer.
How do you differentiate yourself from the rest of the market?
Mark: Our depth and breadth of experience in industry-specific AI-driven speech solutions differentiates us. To put some scale on that, Nuance AI is built into 250 million automobiles and powers 600 million customer service chats each year. We also deliver text-to-speech capability in 125 unique voices for 50 languages. Our technology powers 400 million consumer voiceprints and biometric security systems at the world’s largest financial institutions. Almost 80 percent of all radiologists and more than half of all physicians in North America use our systems daily to manage and document patient care.
The challenges and opportunities in the healthcare market attract large and small players with varied AI and market experience. Conversational AI for healthcare is a much heavier lift than what many entering the market have dealt with in the consumer realm. Medical environments are complex and rely on deeply specialized structured and unstructured medical data and knowledge, as well as human judgment. Speed and accuracy are paramount, and miscommunication can be a matter of life or death.
Deep and specialized datasets to train and refine AI to work in healthcare also are critical. So is having an intimate understanding of the workflows in which AI is deployed. In both cases our solutions leverage our decades of integration with the leading electronic medical record systems that clinicians work in every day, all day. That requires a higher level of vertical specialization and extensive training in the language and practice of medicine.
Then there’s healthcare data privacy and security covered by federal regulations like HIPAA. About 90 percent of hospitals and almost 600,000 physicians rely on Nuance every day. We are not in the business of selling consumer information or using cloud-based data centers for unrelated purposes. Doctors, healthcare systems, and patients also would reject the use of a healthcare solution that might be used for consumer data mining purposes.
How are you going to move forward the AI paradigm in healthcare?
Mark: It’s interesting because if you put 100 technologies in front of clinicians, they’ll say from experience that 95 percent of those technologies make their jobs harder. Instead of simply interacting with the patient and naturally dictating their notes, doctors deal with multiple complex screens and repeated mouse clicks just to properly document and get paid for their services. That gap between promised benefits and real-world experience exists. For example, in specialties like radiology, earlier technologies such as computer-aided diagnosis didn’t live up to initial expectations, so there’s a lot of skepticism about AI.
We think of Conversational AI as a painkiller that removes the pains imposed by administrative requirements and economic realities – allowing clinicians to work more effectively. In healthcare, our Conversational AI lessens the administrative burden for physicians, nurses, and other caregivers, and gives them more time with patients, allowing them to be more present in these interactions. They’re also not staying late or extending the workday at home. It’s striking to see the intense, positive reaction doctors have when they see how this technology can help them. In spite of their skepticism, physicians maintain the hope and motivation to use AI to do their jobs better and to realize their passion for medicine and healing.
Remember, too, that medicine is an evidence-based profession. Physicians want proof that the technologies they’re using are in fact helping them take better care of patients. Likewise, they want to see how these technologies optimize operations and financial results. Increasingly, they’re seeing published results from studies quantifying the benefits of AI. We have measurable clinical and financial outcomes from our physicians and radiologists who also indicate that they are less likely to leave the profession or retire early, and that there is in fact joy in practicing medicine again. That’s what moves the needle.
What's your vision for these technologies and where do you think we're going next?
Mark: The next step in the evolution of Conversational AI can be found in our progress with virtual assistants, and we also are enthusiastic about our recent breakthroughs in a new category of AI-powered healthcare technology: Ambient Clinical Intelligence. The ultimate vision is to reinvent the entire doctor-patient experience in a way that allows clinical documentation to essentially write itself – and help cure physician burnout.
Let’s start with the virtual assistants. Physicians spend a great deal of time getting information into electronic medical records—about two hours of documentation time for every one hour of patient care. Now is the time to change that. With our virtual assistant, we can simplify information retrieval from medical records and empower physicians to use their voice instead of the mouse, giving them easy access to the patient’s history and lab results, or to order prescriptions and tests with conversational, natural voice interactions. We feel so strongly about the benefit of our virtual assistant that we are making it available to our 300,000 Dragon Medical One users in the cloud.
This progress has laid the groundwork for the next breakthrough—our Ambient Clinical Intelligence, which surrounds the care team with knowledge so doctors can focus on the patient instead of the computer. These innovations are still in their early days, but we’re working toward an intelligent exam room that listens to, captures, and understands the conversations in the room, as they happen. Conversations among clinicians, patients, family, and patient advocates are documented in real time, and we apply our intelligence to automatically convert those unstructured human conversations into structured, relevant clinical data within the patient’s health record. Instead of interacting with the computer, the physicians are engaged and listen to the patient, make eye contact, and hold a hand if needed.
When the visit is over, a high-quality draft of the documentation and the key structured data elements are complete and ready for verification. Reinventing the doctor-patient experience in this way will also have downstream benefits outside the exam room, in coding, quality reporting, and compliance for example. At the HIMSS conference this year, we gave people the opportunity to experience it firsthand. It is truly exciting and game-changing.
What if Nuance did not exist? How would healthcare change?
Mark: It’s interesting to gauge your company’s impact by considering its absence like in the film “It’s a Wonderful Life.” The ultimate goal in healthcare is to achieve the quadruple aim: enhance the patient experience; improve the work life of health care providers, clinicians, and staff; improve the quality of care and population health; and reduce costs. Without our technologies and services, progress toward these goals would be hindered.
Any business technology, including AI, should be a workforce productivity tool. In healthcare, it should enable physicians, radiologists, and nurses to do their jobs better, faster, and more accurately. This is what we do, so without Nuance, they would have to revert to manual, time-intensive workflows.
The flip side of that is we’ve been capturing hundreds of millions of patient stories for years, and now we are empowering the physicians to retrieve information with their voice as well. We like to think of AI not in terms of artificial intelligence, but as augmented intelligence. It not only handles the repetitive or low-value reporting tasks that keep physicians from delivering care, but rather boosts providers’ knowledge with a trove of readily available, easily accessed, evidence-based clinical intelligence in real time.
We all read and hear a lot about the benefits that AI can have that enable us to live and work more efficiently. The best and highest-value impact that Nuance can have is to empower the people who care for us. We acknowledge healthcare’s pain – and we are well on our way to finding a cure.
