CerviSense

Real-time posture tracking with instant AI feedback
Key health metrics showing risk, recovery, and posture improvement
Track daily posture sessions, duration, and strain levels
View daily and weekly reports to monitor posture trends
AI analyzes posture trends and performance over time
Insights on correction speed and posture behavior patterns
Challenges and rewards to build consistent posture habits
Personalized posture recommendations based on user data

Inspiration

An AI-driven wearable system that doesn’t just detect posture—it actively corrects and learns from it.

With the rise of digital lifestyles, prolonged screen usage has become unavoidable—especially among students and professionals.

We observed that many people around us frequently experienced neck pain, yet lacked awareness or tools to correct their posture in real time. What begins as a minor discomfort often evolves into long-term cervical issues affecting productivity, well-being, and quality of life.

This motivated us to build CerviSense—an AI for Good solution focused on prevention rather than cure, helping users actively maintain healthy posture and avoid chronic disorders.

What it does

CerviSense is an AI-powered smart wearable that continuously monitors cervical posture and provides real-time corrective feedback. Unlike traditional posture correctors, CerviSense combines real-time sensing, AI-driven decision-making, and active feedback in a single closed-loop system.

Tracks neck angle, tilt, and motion using embedded sensors
Uses machine learning to classify posture as healthy or harmful
Delivers instant vibration feedback for immediate correction
Generates personalized insights and posture trends
Adapts to individual user behavior over time

Detect → Analyze → Decide → Act → Learn

Unlike passive monitoring tools, CerviSense actively intervenes in real time, promoting healthier habits and reducing long-term risk.

How we built it

CerviSense is designed as an end-to-end intelligent health system, integrating hardware, AI, and data-driven insights.

🔹 Hardware Layer

ESP32 microcontroller
MPU6050 IMU sensor (accelerometer + gyroscope)
Coin vibration motor for feedback
Compact, ergonomic wearable patch

🔹 AI & Data Intelligence

CerviSense uses a lightweight machine learning model (Random Forest / Decision Tree) to convert real-time sensor data into intelligent posture decisions.

Instead of relying on fixed thresholds, the system learns from patterns in:

Neck angle deviation
Motion dynamics
Duration of poor posture

The model classifies posture as healthy or harmful in real time (<1s latency), enabling instant corrective feedback.

This allows CerviSense to adapt to individual users, reduce false alerts, and provide more accurate posture correction compared to traditional systems. This transforms CerviSense from a simple monitoring device into an adaptive, intelligent health system.

🔹 Platform Layer

Backend (Supabase) for data storage and APIs
Dashboard for posture analytics and trends
Insight generation for long-term improvement

🔹 Core Logic

We model posture intelligence as:

$$ P = f(\theta, \omega, t) $$

Where posture is continuously analyzed and classified into healthy or harmful states.

if posture == "poor":
    trigger_vibration()

System Workflow

To better illustrate how CerviSense operates in real time, the system workflow is shown below:

[ Wearable Patch ]
(ESP32 + IMU)
        │
        ▼
[ Data Acquisition ] → [ Preprocessing ] → [ AI Model ] → [ Classification ] → [ Feedback ] → [ User Correction ] → [ Analytics  & Insights  ]
   (Angle, Motion)      (Filter + Features)    (Edge ML)        (Good / Poor)      (Vibration)         (Posture Fix)        (Dashboard)
                                                                                                              │
                                                                                                              ▼
                                                                                                   [ Learning Loop ]
                                                                                                   (Adaptive AI)
                                                                                                              │
                                                                                                              └────────────↺ Continuous Improvement

Challenges we ran into

Building CerviSense in real-world conditions introduced several challenges.

One of the biggest hurdles was accurate sensor calibration, as posture varies significantly between individuals. We also had to handle noisy IMU data, requiring filtering techniques to ensure reliable predictions.

Achieving low-latency real-time processing on an edge device (ESP32) required careful optimization of both the model and computation pipeline. At the same time, balancing user comfort with hardware constraints was essential for making the device practical and wearable.

Finally, ensuring battery efficiency and adaptability to diverse posture patterns required continuous iteration.

Accomplishments that we're proud of

CerviSense represents more than just a prototype—it demonstrates a shift toward preventive healthcare.

We successfully developed a functional wearable system capable of real-time posture detection and correction. By integrating AI with edge computing, we created a solution that operates efficiently without relying on external processing.

Most importantly, we built a complete pipeline—from sensing to intelligent feedback to long-term analytics, transforming an idea into a scalable and impactful system.

What we learned

This project taught us that real-world engineering goes far beyond theoretical design.

We gained hands-on experience in IMU-based motion tracking, deploying machine learning on constrained hardware, and designing systems that prioritize user comfort and usability.

Beyond technical skills, we learned the importance of building solutions that are not only innovative, but also practical, accessible, and meaningful for real users.

What's next for CerviSense

CerviSense is just the beginning of a broader vision.

We plan to enhance the system with advanced AI models and larger datasets to improve accuracy and personalization. A dedicated mobile application will provide deeper insights and a seamless user experience.

We also aim to collaborate with physiotherapists and healthcare professionals for clinical validation.

In the long term, we envision CerviSense evolving into a comprehensive posture and musculoskeletal health platform, applicable across workplaces, education, and healthcare ecosystems.