KenYew / parkinson-disease-frequency-feature-engineeering

This repository contains the source code for the frequency domain data analysis of the UCL Final Year Project: Feature Engineering for Detection of Parkinson Disease Severity with Motion Sensor Arrays.

Feature Engineering for Detection of Parkinson Disease Severity with Motion Sensor Arrays

Frequency Domain Data Analysis

Last Modified: 30th Jul 2020
Author: Ken Yew Piong
Department: MEng Electronic Engineering with Computer Science

Description:

This is a standalone tool for frequency domain data analysis with the following features:

1. Data Pre-processing: High Pass Filter to remove gravity component DC offset of accelerometer sensor data
2. Data Pre-processing: Fast Fourier Transform to transform time series sensor data into discrete frequency components
3. Feature Engineering: Extraction of insightful frequency domain features of PD gestures using statistical tools (e.g.: mean, std, iqr, skewness, kurtosis) 
4. Data Visualisation: Visualisation of frequency domain features against different levels of UPDRS rating PD severity