Somerville, Mass.-based Mass General Brigham has been among the earliest, and most aggressive, adopters of artificial intelligence in healthcare.
Becker's recently conducted a Q&A with Nallan Sriraman, chief technology officer of Mass General Brigham, about the AI successes — and challenges — at the $19 billion academic health system.
Question: What is a successful AI project you've been involved with recently at Mass General Brigham?
Nallan Sriraman: We are very proud of our efforts to roll out ambient documentation capabilities, which place the relationship between caregivers and patients at the center of healthcare delivery. Ambient documentation technology securely records patient-clinician conversations and uses AI to generate clinical notes. The clinician then reviews and edits the note before signing the note in the patient's electronic health record.
We started off with 20 clinicians as proof of concept, and that resulted in overwhelming success. That led to an expansion of the pilot, which is now in about 900 clinicians across our system, with an eventual rollout to all our clinicians within the next 24 months.
As chief technology officer, my role in this project is to ensure that our technology partners adhere to the same ethical AI practices we uphold and comply with our data privacy and governance policies. Additionally, I ensure that their technology footprint aligns with and is compatible with our overall technology strategy.
Q: As the chief technology officer, what is your role when it comes to AI (compared to your other C-level colleagues)?
NS: The chief technology officer normally drives the overall AI strategy in collaboration with other CxOs by defining a long-term vision, selecting appropriate technologies, and ensuring seamless integration and scalability.
At Mass General Brigham, our chief information officer, Adam Landman, MD, our chief medical information officer, Rebecca Mishuris, MD, and I co-lead our AI strategy, implementation and operations. Together, we foster interdisciplinary collaboration and external partnerships.
I establish robust data governance policies and define strategies for an effective data management platform. I work with organizational leaders to deploy advanced analytics for actionable insights and provide guidance on ethical AI frameworks to mitigate biases and ensure transparency.
In conjunction with the chief information security officer, Dave Heaney, I also focus on data security and privacy, continuously monitoring AI performance and conducting risk assessments to align AI initiatives with organizational goals and regulatory requirements.
In summary, as CTO for MGB, I drive the technology and data platform strategy that powers AI while collaborating with other C-suite executives to promote the ethical and responsible adoption of AI across the organization.
Q: What is the biggest challenge or obstacle facing healthcare in implementing AI?
NS: The biggest challenge in implementing AI in healthcare has been ensuring patient privacy and security while maintaining compliance with stringent regulations like HIPAA. At the same time, making data available for clinical research is critical. This involves overcoming challenges in deidentification and dealing with the lack of large, high-quality datasets. Additionally, finding cures for rare diseases is particularly challenging because AI relies on large datasets, which are inherently limited for rare conditions.
Balancing these regulatory requirements, ensuring data accuracy, addressing ethical considerations, and gaining the trust of healthcare professionals and patients add significant complexity to developing and deploying AI solutions in this field.
Q: What is unique about AI in healthcare compared to the other industries you've worked in?
NS: AI in healthcare is uniquely challenging due to its heavily regulated environment, the critical importance of data accuracy, privacy, trustworthiness, and profound ethical implications. It requires interdisciplinary collaboration, extensive clinical validation, and a focus on patient-centric outcomes.
Healthcare data's complexity and adoption barriers also set it apart, necessitating a nuanced approach that balances innovation with patient safety and ethical standards.