Designing A More Proactive Continuous Glucose Monitor
Overview
Problem StatementContinuous glucose monitors (CGMs) have transformed diabetes management by providing real-time blood glucose data and reducing reliance on fingerstick testing (SMBG). However, barriers still exist. Users experience pain from needle insertion, emotional distress from “alarm fatigue,” and inconsistent physician adoption due to workflow challenges and patient compliance concerns (Stone et al., 2020).
SolutionCadenceHealth proposes a more proactive CGM system that minimizes patient discomfort and emotional fatigue while improving physician-patient data sharing. The concept emphasizes adaptive alert algorithms and a noninvasive or microinvasive sensor design to enhance usability and long-term adherence.
Process
Market Research and Needsfinding
I began by identifying pain points within the CGM market through literature review and stakeholder interviews. I spoke with a long-term user of the Freestyle Libre, who described the technology as “life-changing,” particularly for identifying trigger foods and regulating blood glucose levels.
From user feedback and published data, I identified key strengths of existing CGMs — real-time feedback, customizable alerts, and intuitive setup — as well as areas for improvement:
-
Pain during needle insertion
-
Emotional burnout from frequent alerts
- Limited clinician integration
Freestyle Libre 3 mobile app dashboard Dexcom CLARITY mobile app user dashboard
Ideation
Using these insights, I brainstormed solutions centered around reducing friction in user experience and increasing device empathy. I explored micro-needle technologies, haptic notifications in place of auditory alarms, and AI-driven trend prediction to alert users before thresholds are crossed.
Low-Fidelity Prototyping
User TestingIn class design sessions, peers and mentors provided usability feedback. One recurring insight was to “make the alert system feel like a guide, not a warning.” This encouraged me to rework the interface tone and notification system to emphasize supportive guidance over anxiety-inducing signals.
High-Fidelity Prototyping
To make the results intuitive, I designed a two-part classification system:
- Effectors – foods that raise the risk of high blood glucose
- Neutralizers – foods that stabilize levels when eaten alongside high-risk foods
Meal History & Tracking
CadenceHealth also includes a bookkeeping feature where users can browse past meals, view associated glucose data, and log repeated meals. Each entry highlights:
- Main ingredients
- Risk levels visualized by color (dark red = high risk → dark green = strong neutralizer)
CaptureAI: Smarter Meal Insights
A flagship feature of CadenceHealth is the integration of machine learning to connect food intake with glucose trends. After a user uploads a photograph of their meal, CaptureAI analyzes the image and generates:
- A breakdown of identified ingredients
- A graph showing glucose response one hour after eating
- A classification of the meal’s food types
Reflection
Takeaways
This project taught me that meaningful medical design goes beyond functionality—it’s about empathy. Through user interviews, I learned that barriers like alarm fatigue and discomfort aren’t just technical problems, but emotional ones.
Designing features like CaptureAI showed me how machine learning can make health data more personal and actionable. Instead of overwhelming users with numbers, it can translate data into guidance that feels human.
Ultimately, this project reinforced my belief that great healthcare technology empowers users—not just to monitor their health, but to trust and engage with it.