by: Peter Hirschmann

Diagnosing Methods  Monitor symptoms such as: Resting Tremor Bradykinesia Rigidity Postural Instability  Sub-symptom Voice Problems  Use classification teaching algorithms to identify Parkinson’s

Parkinson’s Disease  “Movement disorder that is chronic and progressive” Parkinson's Disease Foundation  There is currently no cure  Treatment involves surgery or medication Parkinson Disease

Data – UCI Machine Learning Repository  MDVP:Fo(Hz) - Average vocal fundamental frequency  MDVP:Fhi(Hz) - Maximum vocal fundamental frequency  MDVP:Flo(Hz) - Minimum vocal fundamental frequency  MDVP:Jitter(%),MDVP:Jitter(Abs),MDVP:RAP,MDVP:PPQ,Jitter:DDP – Several measures of variation in fundamental frequency  MDVP:Shimmer,MDVP:Shimmer(dB),Shimmer:APQ3,Shimmer:APQ5,MDVP: APQ,Shimmer:DDA - Several measures of variation in amplitude  NHR,HNR - Two measures of ratio of noise to tonal components in the voice  RPDE,D2 - Two nonlinear dynamical complexity measures  DFA - Signal fractal scaling exponent spread1,spread2,PPE - Three nonlinear measures of fundamental frequency variation  Status - Health status of the subject (one) - Parkinson's, (zero) - healthy  Leave One Out – Useful for realistic testing, since all known data would be used for testing new patients.

Classification Methods

Results – Polynomial Model  Training Error - Blue  Testing Error - Green  LOO Error - Red  Clearly, LOO has the lowest Sum of Square Error Sum Square Error Features 1-22

Results – Maximum Likelihood  Training Data increases with x-axis and Testing Data Decreases Classification Rate  LOO testing method Samples: Samples: 1-195

Results – Nearest Neighbor  7 neighbors, Classification Rate vs. Percentage of Data as Testing Data  LOO Method, Classification Rate vs. # of Neighbors Classification Rate # of Neighbors 1-7 Percentage of Data used as Testing Data 1%-95%

Summary

Conclusion