Kite | Devpost

games

Inspiration

Hospital waiting rooms are emotionally charged spaces, especially in pediatrics and neurology, yet they generate no structured clinical insight.

Children often express anxiety, fatigue, attention differences, or motor challenges while waiting, but these signals are lost before the appointment even begins.

We built Kite to change that: to turn unavoidable waiting time into a quiet, ethical, and clinically useful source of insight without disrupting workflows or adding pressure to patients or clinicians.

What it does

Kite is a gamified stealth assessment platform designed for pediatric clinical waiting rooms.

It engages children in short, playful interactions with shapes, colors, and motion. While they play, Kite captures objective cognitive, motor, and emotional signals and tracks how those signals evolve over time.

From the child’s perspective: it’s just a game

From the clinician’s perspective: it’s structured behavioral measurement

Kite does not diagnose. It observes, tracks, and flags trajectories that may warrant closer attention.

How we built it

We built Kite as a real, end-to-end system, optimized for smooth gameplay, reliable data capture, and clinician usability.

Core Platform:

Mobile App: React Native (Expo) + TypeScript + high-performance rendering for low-latency touch interactions
Backend API: Node.js + Express, Gemini with typed contracts for event ingestion and metric computation
Database: PostgreSQL for structured, longitudinal patient metrics
Dashboard: Web-based clinician view for summaries, trends, and flags

Data Pipeline:

Games emit timestamped raw events (taps, drags, misses, boundary exits)
Events are transformed into structured features:**
Cognitive (reaction time, inhibition, working memory)
Motor (path deviation via RMSE, smoothness, fatigue)
Emotional (non-verbal expression through color and interaction patterns)
Metrics are age-normalized and tracked across visits** Trend-based flags surface meaningful change over time

Clinical Translation One-page Report Scroll with traffic-light indicators Clear, human-readable summaries Dismissible flags to avoid alert fatigue and preserve clinician control

Challenges we ran into

Designing for ethics and safety: We had to ensure Kite surfaces signals, not diagnoses, to avoid overreach and bias.
Balancing engagement with measurement: Too “game-like” risks noisy data; too clinical risks anxiety and disengagement.
Non-verbal emotional modeling: Mapping color choice and interaction patterns to emotional signals required careful framing to avoid oversimplification.
Clinical workflow fit: Any added burden for clinicians would defeat the purpose, so reports had to be minimal, interpretable, and fast.

Accomplishments that we're proud of

Turning a universally negative experience (waiting) into a positive, value-generating moment
Designing an assessment system that works without language
Creating a longitudinal model that captures rate of change, not just snapshots
Keeping the system ethically defensible by avoiding diagnosis and labeling
Building something that feels genuinely fun while producing clinically meaningful data

What we learned

Children and vulnerable patients express far more through interaction than through words
Longitudinal trends often matter more than absolute scores
Designing for trust and safety is just as important as technical accuracy

What's next for Kite

Feature Expansion:

New games to measure sustained attention, task switching, and motor symmetry
Adaptive difficulty to support a wider range of ages and abilities
Exploration of passive physiological signals (e.g., camera-based heart rate estimation, opt-in and privacy-safe) Intelligence & Insights:
Detection of regressions, recovery patterns, and abnormal variability over time
Context-aware baselines (fatigue, inpatient vs outpatient settings)
Optional caregiver-facing summaries for transparency and reassurance Real-World Impact:
Validation of digital biomarkers against established pediatric assessments
Pilot deployments in pediatric clinics and community health centers
Integration with existing clinical workflows and EHR systems