Helping new CPAP users stick with therapy
DURATION
24 weeks/
June-Dec 2024
ROLE
User Experience Design
User Research
User Testing
COLLABORATION
UX Designer
Software Engineer
TL;DR
During my internship at Philips Sleep & Respiratory Care, I worked on improving the early CPAP experience for DreamStation—Philips’ core CPAP device line.
I helped define an early-use experience that makes CPAP feel more comfortable, and less overwhelming for first-time users. To do this, I launched a workshop to align teams on adaptive UI directions, then translated concepts into prototypes.
This work resulted in a validated set of design principles and prototype concepts that shaped a more supportive CPAP experience and established a clear direction for future product development.
Challenge
What is the current problem?
Too Complex to Set Up. Too Hard to Know If It’s Working.
Complex setup: First-time use involves unfamiliar medical terms and multiple device settings, often without clinical guidance at home.
Unclear feedback: Users lack clear signals to know whether the device is working correctly.
Rising competition: Products like ResMed are setting higher expectations for intelligent, user-friendly experiences.
Medical safety constraints: Any intelligent assistance must be predictable, transparent, and non-intrusive.
The previous DreamStation interface - Medical Complexity vs. Cognitive Burden
A user managing sleep therapy with CPAP device
"How might we simplify onboarding for first-time users?"
Research
I began with immersive research to understand how users experience complexity during therapy, what breaks trust or confidence along the way, and what guidance they need to feel supported.
So I took these steps:
User Interviews
I spoke with users and asked them to rank features from most to least important during early therapy use. I find that many functions are important—but rarely used.
User Behaviour Analysis
I mapped how users interact with the device across different therapy moments—before sleep, during therapy, and after waking up—to understand when interactions are intentional versus passive.
Stakeholder Interviews
I interviewed internal clinical and product stakeholders to understand usage patterns, safety constraints, and which settings truly require user control.
Across research, one pattern kept emerging:
Users respond best to support that is subtle, predictable, and easy to ignore when not needed.
Help that demands attention, decisions, or interpretation often increases anxiety—especially in early therapy.
Method
After identifying the structural tension between medical complexity and cognitive burden, the next challenge was translating this insight into concrete, low-risk design directions. Rather than jumping directly into solutions, I facilitated a cross-functional workshop to explore how Adaptive User Interface (AUI) principles could be applied in a high-stakes medical context—while respecting clinical safety, predictability, and user control.
The workshop served as a decision-making tool: to align stakeholders, evaluate trade-offs, and narrow a broad design space into a small set of viable AUI patterns.
01 · Align on what Adaptive UI means
I opened the workshop by framing Adaptive UI specifically for high-stakes medical devices—focusing on gradual adaptation, predictable behavior, and preserving user control.
(Full slides available as a PDF)
02 · Explore good vs. bad Adaptive UI
Participants shared examples of Adaptive UI they trusted—and patterns that felt intrusive, confusing, or unsafe.
This step helped surface shared intuitions around predictability and transparency.
03 · Identify viable AUI patterns
Inspired by the Drunk Island framework, we mapped ideas across impact and implementation difficulty, then converged on patterns that balanced cognitive relief with feasibility and safety.
04 Synthesize and document
After the workshop, I synthesized the discussion into a set of shared design principles and interaction patterns, which became the foundation for subsequent prototyping.
The workshop outputs clarified which AUI patterns were both desirable and viable—directly informing the strategy that followed.
Strategy
Adaptive UI as a low-risk interaction layer
Based on the workshop outcomes, I framed Adaptive UI not as automation, but as a low-risk interaction layer that manages complexity over time.
This strategy followed three principles:
01 Prioritize information by moment of use, not by clinical importance alone.
02 Let intelligence reshape the interface, not interrupt the user.
03 Adapt to behavior without taking control away from patients.
Solution
Grounded in the strategy and workshop insights, I explored four interface concepts for the DreamStation embedded interface—each operationalizing AUI principles at varying levels of guidance, visibility, and user control.
This set of concepts enabled us to compare how Adaptive UI could reduce early cognitive burden while maintaining safety-critical constraints.
Concept Comparison
Concept A offers the strongest balance for early CPAP onboarding—maximizing cognitive relief and confidence while staying clinically predictable—so we advanced Concept A.
Prototypes
Building on Concept 2, I designed a set of interface explorations to show how Adaptive UI could support patients throughout therapy—without increasing cognitive burden or disrupting clinical safety.
The goal was not to make the system smarter, but to make the experience calmer—especially when adherence matters most.
Adaptive card
prioritization
during therapy
Visual tone adapts to time, season, and context
Users are invited to reflect—without being judged
Information is shared across devices, without repeating the same prompts
Conclusion
This project taught me how to design AI in medical contexts—not by adding intelligence, but by carefully shaping when and how it appears.
By working closely with designers, engineers, and clinical partners, we aligned on a low-risk, adaptive approach to supporting early CPAP use.
I’m thankful for the mentorship and cross-functional support that helped me grow as a designer in a medical context.
