See and respond to movement
Live pose landmarks, joint angles, hold timing, repetitions, and rule-based feedback help make each exercise more understandable.
University of Derby research prototype
Rehabilitation that understands how you move.
AI-DRA brings guided upper-limb stroke rehabilitation, live movement feedback, carer involvement, and research monitoring into one accessible platform, using a standard device camera with no wearable sensors required.
Camera-based movement tracking
Survivor and carer portals
Research-ready monitoring
Good reach. Hold your arm steady.
82%Movement accuracy
7 / 10
repetitions
repetitions
University of Derby seed-funded project
Developed within a guided digital health research environment.
Developed within a guided digital health research environment.
No wearable sensors
Designed around the camera already available on common devices.
Designed around the camera already available on common devices.
Connected support
Survivor, carer, and research views within one platform.
Survivor, carer, and research views within one platform.
Meaningful session data
Progress, completion, movement quality, and engagement signals.
Progress, completion, movement quality, and engagement signals.
The rehabilitation gap
Recovery should not become invisible once a survivor returns home.
Home exercise can be difficult to perform consistently when guidance, reassurance, and visibility of progress are limited.
01Exercise at homeOften without immediate feedback on movement quality.
02Progress becomes hard to seeSurvivors and carers rely on memory or isolated appointments.
03Engagement data is lostResearchers and care teams cannot easily understand what happens between sessions.
One connected response
AI-DRA turns a home exercise session into guided, measurable rehabilitation.
The platform is being designed to support the whole rehabilitation relationship, not only the movement on camera.
Keep carers meaningfully connected
Permission-based linking lets carers support routines, review progress, and receive relevant notifications without taking over the survivor's experience.
Make engagement visible for research
A focused monitoring layer can show who is engaging, which exercises are completed, where sessions are abandoned, and how performance changes over time.
A complete rehabilitation ecosystem
Three experiences. One shared picture of progress.
AI-DRA is differentiated by the way its interfaces work together around the survivor, rather than operating as isolated tools.
Clear guidance before, during, and after every session.
The survivor journey is designed to reduce cognitive load, make exercises easier to follow, and turn progress into something visible and encouraging.
- Accessible authentication, dashboard, exercise library, and profile settings.
- Exercise demonstrations, step-by-step preparation, and live camera guidance.
- Repetition counting, hold timing, session history, progress, and reminders.
- Optional voice instructions, sound effects, and reduced-motion preferences.
William CarterSurvivor dashboardThis week
4 / 5
Sessions completed
Next exercise
Target Touch
10 repetitions
Recent activity
Lift & PlaceCompleted
Grasp & Hold82% score
Why AI-DRA stands apart
More than exercise videos. More practical than specialised hardware.
AI-DRA's strongest advantage is the combination of accessible computer vision, connected support, and research-focused data in a single prototype.
01
Uses devices people already have
No wearable sensor setup, charging routine, or specialised camera is required for the intended experience.
02
Responds to movement, not only completion
The system is being built around joint angles, range of motion, holds, repetitions, speed, and movement quality signals.
03
Connects the rehabilitation network
Survivors, carers, and researchers each receive a purpose-built view instead of sharing one generic dashboard.
04
Built to support evaluation
Session, attempt, completion, engagement, and movement records can support quantitative analysis and future usability studies.
| Capability | AI-DRA | Video-only exercise apps | Wearable-based systems |
|---|---|---|---|
| Real-time movement response | ✓ Designed in | Usually limited | Often available |
| Special equipment required | ✓ No wearable required | None | Typically yes |
| Carer connection | ✓ Dedicated portal | Often absent | Varies |
| Research engagement monitoring | ✓ Purpose-built | Basic analytics | Varies by product |
| Functional task orientation | ✓ Included in exercise design | Varies | Varies |
Movement intelligence
From camera landmarks to rehabilitation feedback.
AI-DRA does not need to store every camera frame. The system can translate live movement into smaller, meaningful rehabilitation measures and session records.
Detect body and hand landmarks
MediaPipe-based tracking identifies the relevant shoulder, elbow, wrist, and hand points.
Interpret the exercise state
Joint angles, position thresholds, timing, and movement stages determine whether a repetition is progressing correctly.
Guide the survivor live
On-screen and optional spoken prompts explain when to raise, hold, return, slow down, or adjust.
Summarise the session
Completion, accuracy, speed, range of motion, holds, and attempt states can contribute to a movement score and recommendation.
Session movement analysisPrototype data model
84%
76°
Steady
Movement quality across repetitionsRep 1–8
Completed repetitions8 of 10
Repeated incomplete attempts2 detected
Suggested next stepRepeat at current difficulty
What the prototype contains
A substantial product foundation, not a single demo screen.
The status labels below keep the landing page honest by distinguishing built interface foundations from the deeper movement intelligence still being completed.
Survivor dashboard and navigation
Welcome view, weekly progress, activity cards, quick actions, sessions, profile, history, reminders, and settings.
Exercise library and guidance
Exercise browsing, detail pages, demonstrations, preparation steps, and task-based activities such as Target Touch, Grasp & Hold, and Lift & Place.
Camera movement tracking
Pose landmarks, arm selection, joint-angle calculation, hold timing, repetition counting, camera controls, and rule-based feedback.
Carer invitation and settings flows
Invitation entry, pending acceptance, resend, edit, cancel, linked-survivor concepts, notifications, and profile, password, help, and preference pages.
Research monitoring dashboard
Dashboard, survivors, carers, notifications, exercises, and settings views focused on engagement, completion, abandonment, consistency, and system performance.
Attempt states and movement scoring
Completed, failed, and incomplete attempts; paused-session rules; live accuracy; movement score; compensation flags; summaries; and recommendations.
Research monitoring concept
Engagement made measurable.
The dashboard is designed to support future quantitative evaluation without turning the project into a commercial administration system.
Sessions completed78%
Weekly consistency66%
Exercise completion84%
Primary question
Who is engaging?
Participation frequency and usage patterns across the prototype.
Primary question
What gets completed?
Exercise uptake, session outcomes, and points of abandonment.
Research-led by design
Built to learn from real use, not only to look impressive in a demonstration.
AI-DRA is a seed-funded research prototype. Its purpose is to explore how digital rehabilitation can be accessible, usable, and measurable for stroke survivors and the people supporting them.
Privacy-aware data design
Prioritise meaningful movement and session measures rather than storing unnecessary raw camera data.
Co-design and usability focus
Survivor, carer, and healthcare professional feedback can directly shape wording, interactions, accessibility, and workflow.
Transparent prototype evaluation
Separate validated functionality from features still under development, then test both usability and technical performance.
The survivor journey
Designed to make each session feel clear, supported, and achievable.
The interface reduces unnecessary steps and guides the survivor from preparation to reflection.
01
Choose an exercise
Browse assigned or recommended upper-limb and functional activities.
02
Prepare safely
Review simple instructions, positioning, camera setup, and audio preferences.
03
Move with guidance
Follow visual and optional spoken prompts while the system tracks the exercise.
04
Review progress
See session completion, movement measures, history, and the suggested next step.