BioSyncAI: Closed-Loop Biofeedback for Flow State Maintenance

BioSync AI is a real-time system designed to combat the "Fragmented Attention" and "Dopamine Disparity" caused by modern social media algorithms. By utilizing heart rate variability (HRV) as a proxy for the Central Autonomic Network (CAN), the system detects cognitive states and uses generative music as a regulatory tool to guide users into an optimal "Flow State".

🧠 Project Overview

The core objective of this project is to answer: Can a real-time AI system use physiological biomarkers to detect and regulate a user's cognitive state?. Rather than treating music as static background noise, this system treats it as a dynamic, closed-loop biofeedback signal.

Theoretical Framework

• Flow Theory: Based on Csikszentmihalyi’s definition of flow as a state of complete absorption in the present moment.
• Neurovisceral Integration: Based on Thayer and Lane’s model, where heart rate variability (HRV) reflects the functional integrity of the Central Autonomic Network (CAN).
• The Nine-Quadrant Model: The system classifies user states into categories such as Anxiety, Arousal, Flow, Control, Relaxation, Boredom, Apathy, and Worry.

🛠️ Technical Architecture

1. Data Acquisition (The Input)

The system extracts real-time physiological data through a wearable HRV sensor. Key features include:

• BPM (Beats Per Minute).
• RMSSD (Root Mean Square of Successive Differences).
• HR Delta (Heart Rate variability trends).

2. Intelligence Layer (SVM Classification)

The system utilizes a Multidimensional Multiclass Support Vector Machine (SVM) to predict cognitive states.

• Kernel Trick: Projects biomarkers into higher-dimensional space to handle non-linear physiological data.
• Feature Scaling: Standardizes data to ensure the hyperplane is not biased toward dimensions with larger numerical ranges.
• Pre-training: The model is pre-trained on the WESAD dataset to establish baseline stress vs. relaxation boundaries.
• Personalization: Conducts fine-tuning by collecting user-specific data during tasks and labeling them via a Flow State Scale assessment.

3. Generative Music Engine (Reinforcement Learning)

Rather than simple rule-based mapping, the system uses a Reinforcement Learning (RL) framework.

• Modulation: Adjusts parameters such as tempo, rhythmic density, harmonic complexity, and spectral brightness.
• Reward Signal: The stabilization of HRV balance serves as the intrinsic reward, ensuring the user moves toward the Flow quadrant.

🚀 Getting Started

Prerequisites

• Arduino Uno R3 (for real sensor input)
• Pulse Sensor (Analog, e.g., PulseSensor.com heartbeat sensor)
• USB cable for Arduino connection
• Python 3.10+

Arduino Setup

1. Wire the pulse sensor to Arduino:

   • SIGNAL pin → A0 (Analog pin 0)
   • VCC pin → 5V
   • GND pin → GND

2. Upload the Arduino sketch (save as pulse_sensor.ino):

   const int PULSE_PIN = A0;
   void setup() {
     Serial.begin(9600);
   }
   void loop() {
     int val = analogRead(PULSE_PIN);
     Serial.println(val);  // outputs 0-1023
     delay(10);  // ~100 Hz sampling
   }

3. Find the Arduino serial port:

   • macOS: /dev/cu.usbmodem1411 (or similar in /dev/cu.*)
   • Linux: /dev/ttyUSB0
   • Windows: COM3

Python Version

This project requires Python 3.10 or 3.11 (recommended).

Setting Up Python with pyenv (Recommended)

If you need to install or manage multiple Python versions:

# Install pyenv (macOS)
brew install pyenv

# Add to shell config (~/.zshrc or ~/.bashrc)
echo 'export PATH="$HOME/.pyenv/bin:$PATH"' >> ~/.zshrc
echo 'eval "$(pyenv init -)"' >> ~/.zshrc
source ~/.zshrc

# Install Python 3.11
pyenv install 3.11.0

# Set local Python version for this project
pyenv local 3.11.0

# Verify
python --version

Installation

1. Create and activate virtual environment:

   # Create venv
   python -m venv venv
   
   # Activate (macOS/Linux)
   source venv/bin/activate
   
   # Activate (Windows)
   # venv\Scripts\activate

2. Install dependencies:

   pip install -r requirements.txt
   pip install pyserial  # Required for Arduino communication

Running the Code

# Activate venv first (if not already activated)
source venv/bin/activate

# Simulated sensor mode (default - no hardware needed)
python biofeedback_system.py

# Arduino sensor mode (real hardware)
python biofeedback_system.py --sensor arduino
python biofeedback_system.py --sensor arduino --port /dev/cu.usbmodem1411

Hardware Setup

Wire the pulse sensor to Arduino as described above. The system communicates directly with the Arduino over USB serial — no external ADC needed.

Project Structure

BioSyncAI/
├── biofeedback_system.py   # Main system (supports simulated or Arduino sensor)
├── requirements.txt        # Python dependencies
├── README.md              # This file
├── .gitignore            # Git ignore rules
└── pulse_sensor.ino     # Arduino sketch for pulse sensor

Dependencies

• numpy - Numerical computing
• scikit-learn - SVM implementation
• scipy - Signal processing
• pyserial - Arduino USB serial communication

Implementation Notes

• Edge Cases: The system includes personalized calibration to handle baseline distributions and avoid misclassifying pathological irregularities like arrhythmia.
• Deployment: The user's normalized physiological input is fed into the real-time trained SVM classifier to predict current cognitive states.

📚 References

• Csikszentmihalyi, M. (1997). Finding flow: The psychology of engagement with everyday life.
• Thayer, J. F., & Lane, R. D. (2000). A model of neurovisceral integration in emotion regulation and dysregulation.
• Satani, A., et al. (2025). Modern day high: The neurocognitive impact of social media usage.