Nina
Guiding injured athletes through knee rehab with a wearable, app, and AI assistant
MY ROLES
Phone App UI, Wearable Tester (did not test on others due to COVID)
TEAMMATES
Armel Patanian, Katie Zhang, Sara Pope, & Andrew Hemnes
TOOLS
Miro, Figma, Fusion 360, Zoom, Slack
DATE
March - June 2020
Challenge
How might we help athletes rehabilitate knee injuries while quarantining at home and staying in shape?
When the pandemic hit, several classmates and I used the extra time at home to up our fitness regimens. It didn’t take long before we collectively experienced an onslaught of stress-induced injuries - tightened backs, shin splints, and aching knees. We funneled our frustrations into Nina - an AI-enabled wearable to help quarantining athletes stay in shape while managing and healing knee pain. This project was completed entirely through virtual meetings.
Research
Studies of the Self
We observed and documented our own exercise habits and injuries over the course of a week. Two of us experienced knee pain while running for the first time in a while. Collectively, we ran, surfed, lifted weights, stretched, hiked, practiced yoga, and walked our dogs.
According to PhysicalTherapyAide.org, the most common injuries which require PT are 1) pulled muscles and 2) runner’s knee. Since we wanted to design a wearable for a specific body part, we decided to focus the remainder of the project on a knee device.
Domain Expert Interview
Since we completed this project during quarantine (March - June 2020), our in-person research capabilities were limited. However, we learned about PT office responses to COVID from a phone interview with Linda, a receptionist at Rehab 1 Physical Therapy in St. Louis, Missouri. Rehab 1’s clientele dropped by around 75% during quarantine, so the office started 2 varieties of “telehelp appointments” - video chats or audio calls. Since many of Rehab1’s patients are elderly, they often waste the first 30-40 minutes of their appointments trying to set up the necessary technology. Apart from Zoom struggles, Linda reported that "At-home PT is surprisingly effective - almost all exercises can be completed with structures found around the house, like walls and doorjams."
Existing Technology
We also researched existing biometric wearables. SmartKnee uses a gyroscope-enabled cord to send a constant stream of knee angle data to physical therapists. Reflexion Health uses an engaging avatar named Vera and 3D capture sensors to monitor at-home exercise habits. Mio Sensors are lightweight, non obtrusive, Bluetooth-powered sensors that can precisely measure limb angles. Several posture wearable brands adhere to skin and vibrate whenever wearers start to slouch. DorsaVi Wearable Technology assists physical therapists by analyzing patients’ movement patterns at 200 frames per second. Some lightweight, washable, and comfy smart fabrics can help patients recuperate. Sensorimotor-enabled biometric robots can also help patients by learning about them through continued interaction.
Of these existing technologies, a few ideas stood out. We decided to implement similar gyroscopic sensors as SmartKnee due to their flexibility and unobtrusiveness. For data redundancy, we decided to add GMR (giant magnetoresistance) analog sensor chips, which can calculate distance (and therefore knee angle) based on the the field between two magnets. We also liked the idea of marking pain with some sort of dial feedback mechanism. Inspired by Reflexion Health’s Vera, we personified our smart knee brace with a slightly punny name - Nina.
Persona
Cadence Irving
Age: 25
Work: Accountant
Family: Unmarried, close to her younger siblings
Location: San Fernando Valley, Los Angeles
Character: Responsible, optimistic, go-getter, competitive, active swimmer
Frustrations: knee pain + closures of gym, pool, and PT office due to Covid
Bio
Since the COVID-19 pandemic caused gyms to shut down, Cadence could no longer swim or lifeguard. She turned to running to keep up her fitness. After ordering a new arm-band and pair of leggings online, Cadence was feeling optimistic and even a little bit excited for this new phase of her fitness journey. However, after one week of running daily she began to experience joint pain in her knees. How can she continue to exercise without further injuring herself? How can she heal this injury so that it doesn’t become permanent?
Design
User Flow
With Cadence’s goals of rehabilitation and knowledgeable advice in mind, we created a user flow for our wearable - app system.
Simply put, users would:
Buy the wearable
Download the app
Connect the wearable & app
Choose to use Nina or a real, human physical therapist (we later removed the PT feature to narrow project scope)
Follow along with smartly recommended exercises
Record and analyze pain events
View a personalized recovery timeline
App Design
The app has three main capabilities:
Algorithm-based Exercise Suggestions
Statistics of Pain Events / Knee Flexion
GPS Run Tracking
The side nav’s pain scale lets users log discomfort in-app instead of reaching to tap the knee wearable.
Opening
Onboarding
Side Nav + Pain Log
Statistics
Suggested Exercises
Run Tracking
Wearable Design
Our device has clips to attach to existing knee braces. The white strip on the longer portion of the wearable can be swiped by a finger to rate extent of pain and send a report to the Nina app. Sensors embedded in the wearable constantly measure knee angle, so app users can cross-examine pain events with knee flexion.
Low - Fidelity Mockups
Magnetometers
By measuring the field strength between two magnets (one above the knee, one below the knee), we can constantly measure knee flexion.
Strapping In
At first, we thought athletes could attach the device through two straps.
Packaging
The device would arrive with instructions and a unique code to pair with the app.
A Test Run
I tested our primary mockup by trying it on for a run. Turns out, straps weren’t sufficient to hold the wearable in place (see the image on the right), so we moved toward compression sleeves for future iterations.
On this run, we also tried out the Nike Run Club app, which influenced our GPS Go UX design.
Reflections
Questions to Answer with Future Testing
Can our device detect knee flexion with the precision required to cross-analyze against pain events?
Beyond degree of flexion, what other factors contribute to knee pain? How might our device track these (such as speed, force, and sudden movement)?
What is the most intuitive and accurate way for users to report knee pain? Does a slideable scale from yellow to red make sense?
Do people trust algorithmically-suggested exercise videos to carry out the typical role of a physical therapist?