Parkinson's Disease Prediction Model

This feature has been undertaken to predict the presence of Parkinson's Disease based on the features expressed by patients that can be extracted from the patient's vocal patterns. Parkinson's Disease is a neurodegenerative disorder that has the common symptoms of¹:

• Tremors,
• Bradykinesia (Slowed movemend)
• Stiff or rigid muscles
• Impaired posture, balance
• Impaired automatic movements
• Speech changes
• Dexterity degeneration, particularly noticeable in hand writing

At present, the cause of Parkinson's is utterly unknown but more data collected about patients as well as early intervention and care may aid to slow the onset.

Data Analysis

To analyse this data, as well as to train machine learning classification models, a number of python libraries have been incorporated into the project:

• Data Analytics / Transformation:
  • Pandas
  • Numpy
• Visualization:
  • MatPlotLib
  • Seaborn
• Machine Learning:
  • Scikit-Learn

Dataset

The dataset used at this stage is the Oxford Parkinson's Disease Detection Dataset². This dataset consists of 195 voice recordings of 31 people, 23 of which have been diagnosed with Parkinson's Diseas

The data has the following features:

Name	Description	type
Name	An ID denoting a patient	nominal
MDVP:Fo	Average Vocal fundamental frequency	Continuous: [Hz]
MDVP:Fhi	Max. Vocal fundamental frequency	Continuous: [Hz]
MDVP:Flo	Min Vocal fundamental frequency	Continuous:[Hz]
MDVP:Jitter	measure of variation in fundamental frequency	Continuous:[%]
MDVP:Jitter	measure of variation in fundamental frequency	Continuous:[Absolute Value]
MDVP:RAP	measure of variation in fundamental frequency	Continuous
MDVP:PPQ	measure of variation in fundamental frequency	Continuous
Jitter:DDP	measure of variation in fundamental frequency	Continuous
MDVP:Shimmer	measure of variation in amplitude	Continuous
MDVP:Shimmer	measure of variation in amplitude	Continuous: [dB]
Shimmer:APQ3	measure of variation in amplitude	Continuous
Shimmer:APQ5	measure of variation in amplitude	Continuous
MDVP:APQ	measure of variation in amplitude	Continuous
Shimmer:DDA	measure of variation in amplitude	Continuous
NHR	measure of ratio of noise to tonal components in the voice	Continous
HNR	measure of ratio of noise to tonal components in the voice	Continuous
status	Health status of the subject	Nominal: [1: Parkinson's. 0: No Parkinson's]
RPDE	Nonlinear Dynamical Complexity Measure	Continuous
DFA	Signal fractal scaling exponent	Continuous
spread1	Nonlinear measure of fundamental frequency variation	Continuous
spread2	Nonlinear measure of fundamental frequency variation	Continous
D2	Nonlinear measure of fundamental frequency variation	Continuous
PPE		Continuous

Footnotes

1. https://www.mayoclinic.org/diseases-conditions/parkinsons-disease/symptoms-causes/syc-20376055 ↩

2. Little, M. (2007). Parkinsons [Dataset]. UCI Machine Learning Repository. https://doi.org/10.24432/C59C74. Accessible from http://archive.ics.uci.edu/dataset/174/parkinsons ↩