Encouraging patients to take their pills on time and follow their treatment plan can often be a real challenge for both the patients and the doctors monitoring the patients' progress. This is especially true for people living with tuberculosis (TB), where missing medication doses can have serious consequences on their recovery. The challenge is even greater in remote or underserved communities around the world, where access to care and basic technology is minimal. That’s where behavioral health interventions (BHIs) come in. BHIs are personalized support programs that use digital tools, like patients’ smartphones, to send tailored reminders, deliver health tips, and check in with patients during their treatment. The goal? Keep patients on track and help them stick to their treatment. But here’s the issue: most BHIs still use a one-size-fits-all system that treats every patient the same, regardless of their unique needs or special circumstances.
That’s where Professor Justin Boutilier at the Telfer School of Management steps in. With a Discovery Grant from NSERC, Boutilier is building a robust analytics system that could transform how interventions are designed and delivered.
This project relies on strong partnerships with frontline organizations, including Keheala in Kenya, which operates a digital service to promote medication adherence among patients undergoing TB treatment, and NanoHealth in India, a community health worker network focused on screening and managing diabetes in underserved populations.
So how does Boutilier’s analytics system work? By combining machine learning with optimization tools, the research team aims to tailor support for each patient based on their unique conditions, preferences, location, and life circumstances. The system analyzes real-world data from past patients to predict how individuals are likely to respond to different types of support, like a text reminder, a phone call, or an in-person visit. It then figures out the best way and time to deliver that support. This technology can help clinics use their limited financial and human resources more wisely, while ensuring that even patients in low-income and hard-to-reach communities aren’t left behind.
The result? Delivering the right support, to the right patient, at the right time. As Boutilier explains, “My research suggests that personalizing treatment support allows us to use limited resources more effectively, thereby improving individual outcomes and overall program cost-effectiveness. For example, in tuberculosis treatment, the benefits of personalization are highest in situations where resources are most constrained, which is the reality in most high TB burden countries.”
The implications of this work could be transformative for health care delivery.
The World Health Organization has called for better person-centered care through digital health tools, especially in countries where resources are scarce. By moving beyond simply tracking patient behavior to building digital tools that act on that data, Boutilier’s project aims to reduce healthcare costs and deliver targeted support to those who need it most, helping patients stick to their medication and take charge of their health.
What’s more, Boutilier is helping train the next generation of data scientists ready to lead the next wave of ethical, AI-powered solutions in healthcare. Because in the end, it’s not just about reminders, it’s about building a smarter, more equitable health system that meets patients where they are, empowers them to take control of their health, and motivates them to stay on track with their recovery.
