BoneScope
How might we support EMTs in making fracture assessment with the wearable product in emergency situations?
Assessment Interface for Emergency Care
Impact
Improving task completion rate of bone fracture assessment and reducing cognitive load by ~70% among 5+ clinicians.
Increasing ML data labeling accuracy by 26%.
High-fi prototype and task flows delivered in 4 weeks, securing 100% stakeholder approval and alignment.
Role
Product Design Lead, UX Researcher
Apr 2025 - Present(involved from project initiation)
Being the only designer of the team, I own the design process of this product and the design system. I report directly to CEO, Dr. Tim Burdick.
Challenge
Develop a mobile-based medical app that -
guides step-by-step bone assessment procedures fast and with minimal cognitive load,
prepares analysis and medical reports for hospitals,
controls remote device with history records,
supports ML model training with data labeling,
Context
BoneScope is a medical wearable product using digital tuning fork to detect fractured bone without x-rays and has received funding awards from the Dartmouth Digital Health Accelerator 2025.
Moving into the development phase, the device needs a companion smartphone software that monitors and guides a complete assessment process, generates medical reports, and records patient data to continuously train the ML model, enabling more accurate predictions of fracture likelihood over time.
Team: Dr. Tim Burdick, Indrani Bhattacharya, Reet Kothari, Thomas Robinson, Stephen Adjei, Cindy Liu Jiayi
Background
I joined the Dartmouth Digital Health Accelerator 2025 as the sole designer on a 5-person team (2 clinicians, 2 engineers, and myself).
Building on Dr. Burdick’s sound-based prototype for fracture detection, we advanced the concept into a Go-To-Market product through journey mapping, pain point analysis, Pugh matrix evaluation, storyboarding, and test plan mapping.
I initiated the idea of a smartphone app as BoneScope’s digital “tuning fork,” leading early design exploration.
Our team’s work earned the $5,000 pitch competition award.
Preliminary Research
With 3 clinicians, 2 engineers, and 1 business advisor in our team, I led 3 Zoom meetings to understand product goals, user pain points, and market/clinical requirements.
3 Focus Group Sessions
I had conversational interviews with 2 clinicians for medical procedure walk-through, building up a task flow for bone assessment in line with clinical standards.
2 Field Studies
6 Insights
Fast, low-load workflow
Task flow must be quick and minimize cognitive effort.
Seamless sharing
Enable secure transfer of records and analyses to hospitals.
Clear, actionable data
Metrics and visualizations should be obvious and directly support decision-making.
User reassurance
Design should instill confidence and calm for EMTs in high-pressure scenarios.
State clarity
Interface must distinguish device on/off, connection control, and start/end of assessment.
ML-ready records
Input/output structured for continuous model training.
Early Ideation
How might we
Create a bone assessment task flow that is fast and low-cognitive.Minimize mandatory screens
I cut down the procedure steps to:
Start - Info Input - Device Action - Info Output
which in our task flow became:
Start - Patient Info - Scan - Result - Complete
Minimize inputs
To achieve recognition over recall, I designed selection-based input for date of birth, weight, and height with automatic unit conversion.
How might we
Allow the process records both available for hospital and ML model trainingReport page
I created a report page aside from the result to distinguish items available for printing and sharing to the hospital.
This allowed putting most necessary info first, while keeping clinical ready analysis optional to view.
Case-based labeling
The records are saved case by case instead of by patient. This way, not only do we protect patient privacy, we can also learn from assessments that are interrupted.
How might we
Reassure users with clear and actionable data. Other than the mandatory progress bar, I added time left and real-time signal strength, which ensures EMTs are always updated with the current status.
I also created consistent escape hatches including pause and back to last step to give EMTs more control in case of an unhappy path.
I created this wireframe with 2 underlying user research questions for the next iteration:
Do users prefer instructions in the app? If so, what content and how?
Would having patient history or scan records saved in the app be valuable to users? If so, what content and how long to be retained?
Iterative User Research
I conducted user interviews combining moderated usability testing, A/B concept testing, think-aloud, and cognitive walkthroughs.
The interviewees refined the app’s wording to be more medically professional, shared their routine bone-assessment workflow, and confirmed the importance of instruction page and history records.
By observing their navigation and think-aloud feedback, I also captured confusion, points of friction, and preferences for potential features, leading my next step prototype design.
part of my interview notes
Iterative Design
Prototype 1: Iterations from Wireframe
Workflow
