This series introduces D^3 Associates Program projects which aim to answer important questions at the intersection of artificial intelligence and digital technologies in business and society.
This article shares insights from Dr. Susanna Gallani, Tai Family Associate Professor of Business Administration at Harvard Business School, Dr. Lidia Moura, Associate Professor of Neurology at Harvard Medical School, and Dr. Moura Junior, AI researcher at the Mass General Brigham (MGB) Center for AI and Biomedical Informatics for the Learning Healthcare System (CAIBILS) and Research Associate at Harvard Business School, who are pursing research on the topics of artificial intelligence and healthcare.
1. What drew you to this area of research and how did you first become involved in this work?
This project was born from a shared commitment to improving healthcare delivery through innovation. Our multidisciplinary team brings together decades of expertise from different corners of the field. Dr. Susanna Gallani, a management controls scholar, studies how job design and organizational performance affect clinician well-being in healthcare organizations. She sees AI as a powerful lever to help enable value-based care transformation at scale and improve workers’ well-being and effectiveness. Dr. Lidia Moura, a neurologist and neurophysiologist with training in population health sciences and clinical effectiveness, was drawn to AI after facing persistent challenges with data quality in health services research. As she witnessed firsthand the toll of rising burnout and overwhelming patient-provider messaging, she became a strong advocate for leveraging AI to support more efficient, meaningful communication in clinical care. And Dr. Moura Junior, a computer scientist and AI researcher with over 25 years of experience in biomedical informatics and data engineering, focuses on designing scalable solutions that improve healthcare quality, safety, and efficiency through AI and data science. Together, our perspectives intersect around a shared goal: using AI thoughtfully to strengthen the connection between patients and their care teams to deliver better care in a thriving work environment.
2. What are some common misconceptions or barriers around the problem you’re working to solve?
One of the most common misconceptions is the belief that generative AI cannot handle the complexity or nuance required for patient-provider communication – a task often seen as too high-stakes, too sensitive, and too “human.” While it is true that AI may not replace all healthcare provider tasks, the reality is that message volume has grown exponentially, while provider capacity has not. The strain on frontline clinicians is unsustainable, and innovation is essential – not to replace them, but to support them.
Another concern we frequently encounter is the perceived risk of using generative AI in patient-facing contexts. We take these concerns seriously. That is why our team is rigorously evaluating not only the accuracy and effectiveness of AI-generated draft messages, but also their tone, clarity, and even empathy – benchmarking them against human-generated messages. Our current work is not about replacement; it is about developing and validating a proof-of-concept that shows how AI can be a safe, responsible, and scalable part of the solution to a critical challenge in healthcare.
3. What research is being done on this topic and how is your approach or perspective unique?
There is a growing body of research on AI and generative AI applications in healthcare, particularly in areas like clinical documentation. For example, ambient AI scribes – technologies that transcribe and structure clinical conversations – have shown promise in improving documentation efficiency, clinician productivity, and reducing burnout.
However, far less research has focused on AI-generated responses to patient messages – a challenge area that has rapidly grown in complexity. That is where our work stands out. Our approach leverages retrieval-augmented generation (RAG) to incorporate relevant chart context and standard operating procedures (SOP), ensuring messages are accurate and personalized. We have also developed a highly curated, subspecialty-specific dataset of human-benchmark responses and embedded institutional workflows, business logic, and domain knowledge into the model.
Unlike most solutions that focus on administrative support or primary care, our work is tailored to high-complexity subspecialties and grounded in real-world message data. This makes our approach not only novel but also highly applicable to the nuanced communication needs of specialty care.
4. What excites you most about this work and its potential impact?
What excites us most is the truly collaborative nature of this project. It brings together a multidisciplinary team from Harvard Medical School, the Harvard T.H. Chan School of Public Health, and Harvard Business School. Each week, we engage in dynamic discussions that reflect diverse perspectives – from data science and AI to clinical practice and organizational strategy.
Importantly, this is not a solution built in isolation. We have intentionally designed it with input from across the healthcare ecosystem, including MDs, nurses, advanced practice providers, trainees, and patients themselves. That inclusive approach ensures the tool is not only technically sound but also practical, useful, and aligned with real-world needs.
The potential impact is significant. By supporting more efficient, accurate, timely, and compassionate communication, this work can help reduce provider burnout and improve patient care and patient experience – some of the most urgent challenges facing healthcare systems globally.
5. How do you hope working with D^3 will amplify the impact of your work?
We are incredibly grateful for D^3’s vote of confidence and support – especially at a time when securing resources for early-stage research and proof-of-concept development is increasingly competitive. Being selected as the only healthcare-focused project in this cohort is both an honor and a unique opportunity.
What excites us most about this collaboration is the chance to connect with innovators across sectors and to learn from diverse approaches to AI application. The opportunity to exchange of ideas, insights, and feedback from fellow awardees is invaluable. We believe that this kind of cross-pollination – between disciplines, industries, and perspectives – will not only strengthen our project but also help us think more broadly about scalability, implementation, and long-term impact.
6. What changes do you hope to see in your field as a result of the work being done in this area?
We hope this work can serve as a model for what is possible when multidisciplinary teams, informed by real-world input from both patients and providers, come together to co-create solutions. Through our collaboration across multiple Harvard institutions, we aim to show that AI can be both effective and safe in enhancing patient-provider communication.
Ultimately, our goal is to improve patient experience and outcomes by enabling more timely, personalized care – while also easing the burden on healthcare providers. Much of today’s burnout stems from non-clinical work done outside of patient visits, and this project offers a path to reduce that burden.
With broader adoption, we hope this work contributes to the transformation of healthcare into a more sustainable, resilient system – one that better supports both the people who give care and those who receive it.
7. What’s an essential area in which AI and digital technologies will reshape the way businesses or society operate in the long run that we may not be considering?
One critical area that deserves more attention is data privacy and cross-sector data sharing. Despite the transformative potential of generative AI in healthcare, progress is often hindered by limited access to enabling APIs and a deep-rooted hesitation around data use and trust. These concerns – while valid – underscore the need for new models of responsible data stewardship that can foster collaboration across institutions and industries.
Today, many healthcare systems still operate in silos, often dictated by the boundaries of major electronic health record platforms. As a result, even the most promising AI innovations struggle to scale. In the long run, creating frameworks for secure, ethical, and multi-directional data and platform sharing – guided by public trust and shared purpose – could unlock enormous value. Other industries have already begun to realize these gains. Healthcare must follow suit if we hope to realize the full potential of AI to improve outcomes and efficiency at scale.
The D^3 Associates Program supports and accelerates faculty research into the ways AI and digital technologies are reshaping companies, organizations, society, and practice.