The launch of ChatGPT marked a watershed moment in the evolution of Generative AI, catapulting it to the forefront of our collective imagination. Several applications have emerged, from generating marketing content to writing code, enhancing creative potential, and boosting productivity. Healthcare is one of the industries in the spotlight due to AI's transformative power to change it. As tech giants race to revolutionize healthcare with AI, we've seen notable developments such as Google's Med-PaLM-2 and AWS's HealthScribe service, each promising to streamline facets of medical care.
Microsoft has been at the forefront of this revolution, first by acquiring Nuance for $19 billion in 2021 and then launching Dragon Ambient eXperience (DAX) Copilot on Oct 10, 2023, in general availability. DAX Copilot is an AI-powered, voice-enabled solution designed to relieve clinicians from administrative burdens by automatically documenting patient encounters accurately and efficiently right at the point of care.
We spoke with Prerak Garg, Senior Director of Cloud & AI Strategy at Microsoft, who has been the driving force behind the launch of DAX Copilot, to understand the potential of this innovation and, more broadly, of generative AI in healthcare.
Prerak has contributed to shaping the future of digital innovation for some of the world's most influential companies. Prior to Microsoft, he held the position of Engagement Manager at McKinsey, where he worked with Fortune 500 companies to define their digital transformation strategies and implemented Industry 4.0 solutions.
Prerak Garg's technology career began at an unexpected place: Cipla, a global leader in generic pharmaceuticals. As Head of Supply Chain Transformation, Prerak pioneered groundbreaking machine learning solutions at Cipla, revolutionizing supply chain operations. Prominent companies like Meesho, Myntra, Swiggy, and CARS24 implemented similar strategies, drawing inspiration from his work.
Interview With Prerak Garg
Prerak, there's a lot of buzz about DAX Copilot. Can you share what lit the spark for this innovation?
Prerak: Absolutely. You see, it's not just about technology—it's about people. Every day, doctors wrap up their shifts and then face hours of documenting patient visits. It's exhausting. We're talking about a 42% burnout rate in the field, and it's a real problem. The existing fix is to hire human scribes. But they don't come cheap, and not everyone can shell out $60K a year for that kind of help. So, we took a step back and asked, 'How can we make life easier for these doctors?' That's how DAX Copilot came into the picture, shouldering some of the administrative burden on physicians so they can focus on what they do best—caring for patients.
The concept of DAX Copilot is fascinating. How did it evolve from idea to reality?
Prerak: Well, it's a bit like putting together a puzzle. We started with a clear picture in mind: a voice-enabled solution that transcribes patient-physician conversations into medical notes using generative AI. The idea was to have AI step in to reduce the manual effort and cost.
But the real magic happened when we brought different minds together to fine-tune this vision. Collaborative efforts across teams helped us refine the technology to meet high standards of performance and accuracy. We wanted to make it as affordable as possible. This necessitated a horizontal SaaS solution approach, as opposed to the bespoke solutions that are more common in the healthcare industry.
How does your past work experience come into play with DAX Copilot?
Prerak: Oh, it was crucial, to say the least. You see, back at Cipla, I was neck-deep in figuring out how to streamline a very complex process—imagine manually managing the order-to-delivery process across 160 countries! It was daunting. I developed a tool called IMPACT that automated the manual grunt work for our planners, and thanks to machine learning, we could proactively anticipate and resolve supply chain issues. Now, with DAX Copilot, it's a similar breakthrough, but in healthcare. We're using AI to take over the repetitive tasks for physicians.
What is the broader impact of this innovation on healthcare?
Prerak: DAX Copilot is pioneering the shift to cloud-based AI solutions in healthcare. Healthcare has historically been slow to transition to the cloud, primarily due to regulatory challenges and a lack of clear use cases. DAX Copilot demonstrates the value and efficiency of cloud technologies. The impact on the US economy could be even broader: According to National Bureau estimates, widespread adoption of such technology could lead to savings of approximately $200-300 billion annually for the U.S. healthcare system.
So, Prerak, where is generative AI taking us in the world of healthcare?
Prerak: It's like we're at the dawn of a new era—kind of like when Edison flipped the switch on the first light bulb. We're just starting to see how bright the future can be with generative AI in the mix. It will serve as a co-pilot for providers in every aspect of their work. Its application will streamline administrative tasks, improve patient outcomes, and enhance the overall efficiency of the healthcare system.
And it's not just the day-to-day stuff. It can be used in medical imaging for generating synthetic images to train and validate machine-learning models. In drug discovery, it can help generate virtual compounds and molecules with desired properties. It can help automate the tasks of drafting clinical trial communications and translating them into different languages. The possibilities are limitless.
What are some of the cautions of using Generative AI in healthcare?
Prerak: We are still in the early stages of Generative AI in healthcare and must proceed with care. It isn't perfect, and personally, I wouldn't delegate diagnostic responsibilities to it just yet. I'm hopeful it will reach that level of reliability, but we're not there. One of the big issues is 'hallucination,' where AI might give a wrong answer or make stuff up. We need to advance the technology and rigorously test it to ensure it can be trusted in critical healthcare applications without humans in the loop. Currently, I view generative AI as a co-pilot for healthcare professionals, a tool that supports them by enhancing productivity. It's not the primary decision-maker, especially in areas as critical as diagnoses, but it can aid decision-making. Still, it's an invaluable assistant for streamlining tasks and managing information, which allows caregivers to focus more on their patients.
