This project aims to detect and classify abnormal hand movements in Parkinson’s patients using data collected from the onboard accelerometer/gyroscope. The primary goals are:

• Detect tremors (3–5 Hz): rhythmic oscillations due to low dopamine (“Off” state)
• Detect dyskinesia (5–7 Hz): excessive movements caused by excess dopamine (“Too On” state)
• Quantify movement intensity
• Indicate conditions using on-board indicators (e.g., LEDs)

• 📊 3-second sampling window of motion data
• ⚡ FFT-based frequency analysis for condition detection
• 🔔 Condition indication using onboard LEDs
• 🌟 Intensity quantification: Yellow LED (PC_9) lights up when motion intensity exceeds a threshold
• 🔋 Runs on portable power (e.g., power bank)
• 🧩 Implemented entirely on dev board using PlatformIO
• 🚫 No external hardware or modules required

• The onboard accelerometer samples motion data at a fixed rate of 104 Hz.
• A circular buffer stores the last 312 samples (approximately 3 seconds of data).
• For each sample, the magnitude of acceleration is calculated as sqrt(x^2 + y^2 + z^2) to combine the X, Y, and Z axis data.

• A Fast Fourier Transform (FFT) is performed on the most recent 256 samples from the circular buffer.
• The FFT converts the time-domain signal into the frequency domain, allowing analysis of motion frequencies.
• Only the first half of the FFT output is analyzed, as the second half is symmetric for real-valued input signals.
• The frequency resolution is calculated as frequency_resolution = sampling_rate / sample_size, allowing each FFT bin to correspond to a specific frequency.

• The program identifies two types of motion abnormalities based on frequency and amplitude thresholds:
  • Tremor: Detected in the 3–5 Hz frequency range if the amplitude exceeds 14.0 and at least 2 bins meet this condition.
  • Dyskinesia: Detected in the 5–7 Hz frequency range if the amplitude exceeds 15.0 and at least 3 bins meet this condition.

• The system calculates the peak amplitude in both the Tremor (3–5 Hz) and Dyskinesia (5–7 Hz) frequency ranges.
• If the peak amplitude exceeds a predefined threshold (e.g., 100.0), the yellow LED (PC_9) is turned ON to indicate strong motion intensity.
• This feature provides real-time feedback on the severity of the detected symptoms.

• Detection results are indicated using onboard LEDs:
  • Tremor: Only the PB_14 LED is turned ON.
  • Dyskinesia: Both PB_14 LED and PA_5 LED are turned ON.
  • Strong Motion Intensity: The PC_9 LED (yellow) is turned ON when the motion intensity exceeds the threshold.
  • No abnormal motion: All LEDs are turned OFF.

• The detection process is repeated continuously, with a short delay of 10 ms between iterations to maintain real-time responsiveness.
• The system operates entirely on the embedded platform, with no reliance on external hardware or serial output for condition indication.

• Abby Zhang
• Moulik Shah
• Neil Noronha
• Mike Zeng
• Dongting