SPH: You applied your MPH in a unique way—building a career in the business of commercializing products and technology for healthcare at scale. How did your education at SPH prepare you for your chosen path?

Honour: People sometimes ask why I didn’t get an MBA instead of an MPH [but] I’m proud of my MPH because it gave me that macro public health perspective. I zeroed in on epidemiology and biostatistics—I never wanted to be a PhD epidemiologist—but that data training and ability to do research and analyze outcomes served as a foundation for my career. Coming from a clinical family, you don’t always know about the research side, the population side, the business side. It was great to be exposed to that [in undergrad] and then use the MPH as a steppingstone to learn even more.

Perhaps a legacy of my time in the MPH program, I am proud to continue to contribute to peer-reviewed research. For example, I co-authored a publication in Health Affairs on overcoming barriers to adopting and implementing computerized physician order entry systems in U.S. hospitals. Conducting research has provided the foundation for me to be more effective in my professional career evaluating the impact of technology on health outcomes.

SPH: From your experience, what positive impacts is AI having on healthcare, and what are your concerns about the technology?

Honour: AI is a game changer, especially in healthcare. Compared to five or ten years ago, we now have vast amounts of data that can be used to train large language models to quickly answer questions and predict healthcare outcomes. It’s changing how consumers, providers, payers, and pharma researchers make decisions. For example, five years ago, understanding in oncology whether a tumor had risk for metastasis required physician review of images and clinical records. Today, an AI algorithm can review pathology and radiology images and the clinical record in seconds to identify risk.
 
It’s critical that we have guardrails and governance on implementing AI. It’s not going to replace physicians—it enables them to operate at the top of their license so they can take what the AI proposes, the evidence and reasoning, and validate it. Governance and understanding how large language models are trained and the data used is important because hallucinations can occur depending on the data. For public health and healthcare in general, we need to understand inherent biases in models and the data used.

SPH: Where do you see AI having the greatest impact in healthcare?

Honour: There are two major areas where AI will have a huge impact. First, healthcare costs: about 30% of healthcare costs are driven by administrative inefficiencies. For example, it takes multiple manual steps across parts of an organization to determine whether a medication is covered. AI can significantly improve operational processes and reduce administrative inefficiencies, such as in call centers where agents currently review documents manually. That’s low-hanging fruit and very exciting.
 
Second, consumer-facing tools—OpenAI, Anthropic, and Google have healthcare chatbots where consumers can input symptoms and get a response. That spurs innovation in commercial products, but we need governance, so consumers are empowered to engage with providers and insurers with the right questions. Models must be traceable, and healthcare companies must ensure humans review AI outputs. Overall, I think AI will reduce administrative costs and improve outcomes if implemented responsibly.
 
AI can also be used to reach more people. For example, primary care physicians often manage patients with high cholesterol to reduce cardiac events. There’s a disproportionate incidence of cardiac events in some ethnic communities, and patient education is often not tailored to Spanish-speaking or Cambodian-speaking communities. AI agents can be used to create agents that engage patients in a way [specific] to those populations, so the education and interventions are relevant. That’s exciting because you can do culturally tailored engagement at scale in ways you couldn’t before.

SPH: What would you recommend to others seeking to stay abreast of developments in AI?

Honour: My advice is to use it, use it, use it. Force yourself to use it in personal and professional life. There are many free introductory training courses that give basics of AI in healthcare. A basic overview will get people started, but the best way to learn is to start using it yourself. My teams have mandates to integrate AI into every step. We trained product managers to use ChatGPT to build product requirements documents, created and trained agents, and taught skills for teams to get started.