Parkinson's Monitoring · Offline AI

Tremor detection for everyone.

A $25 wrist-worn device that continuously monitors Parkinson's tremors using on-device AI — no smartphone, no WiFi, no infrastructure required.

TremoTrack v1
10M+
People with Parkinson's globally
$25
Total hardware cost
100%
Offline — zero connectivity needed
24/7
Continuous real-world monitoring
How It Works

Simple. Offline. Accurate.

Three steps from wrist to doctor's office — no apps, no cloud, no complexity.

01 — Capture
📡

Sense Movement

An accelerometer and gyroscope sample hand movement 100 times per second, capturing the precise frequency and pattern of every motion throughout the day.

02 — Classify
🧠

On-Device AI

A compressed neural network running directly on the ESP32 microcontroller classifies each movement in real time — distinguishing Parkinson's tremor from normal activity without any internet connection.

03 — Log
💾

SD Card Data

Timestamped tremor events are written to a micro SD card. At the next clinic visit, the patient hands the doctor weeks of continuous real-world data instead of a five-minute observation.

The Problem

Every existing solution assumes connectivity.

We looked at every Parkinson's wearable on the market. Every single one requires a smartphone or WiFi. That means the patients who need monitoring most are the ones who can't use it.

TremoTrack is different.

Built from the ground up for zero-infrastructure environments. The AI runs on the chip itself. The data stays on the card. The patient controls everything.

  • MicrocontrollerESP32 (on-device AI)
  • SensorMPU-6050 IMU
  • AI FrameworkTensorFlow Lite
  • ConnectivityNone required
  • Data StorageMicro SD card
  • Total CostUnder $25

Built by a high schooler who noticed the gap.

I'm a 10th grade student in Apex, North Carolina. I started TremoTrack after researching every existing Parkinson's wearable and finding the same blind spot in all of them — they all assumed the patient had a smartphone and reliable internet.

Most Parkinson's patients don't. So I built something that works for them instead.

TremoTrack is currently in active prototype development. The hardware is assembled, the sensor is communicating, and the AI model is in training. I'm working toward a functional proof-of-concept and plan to validate with local Parkinson's support groups in the Triangle area of NC.

AC
Aarav Chouhan Founder · Green Level High School · Apex, NC
Live Sensor Simulation — TremoTrack v1
ACCEL-X
ACCEL-Y
GYRO-Z
Monitoring Active Recording
4-6 Hz Tremor Band
No Tremor AI Classification
SD Card Logging Data