Cole, minimum local error rate by DNN. Angeles, Paolo,

Cole, Bryan T., et al. 6
, in their paper, examined dynamical machine-learning algorithms, designed to
track the presence and severity of tremor and Dyskinesia with 1-s resolution by
analyzing signals  collected from Parkinson’s
disease (PD) patients wearing small numbers of hybrid sensors. They used  dynamic neural network, and dynamic SVMs,
Hidden Markov Models ,to track the tremors and dyskinesia, on data is collected
from wearable sensors. From different positions of work like sitting, standing,
walking, different models were applied with minimum global error rate showed by
HMM, and minimum local error rate by DNN.
Angeles, Paolo, et al. 7 have developed a sensor system to record kinetic data
from the arm during physical assessments of Parkinson’s disease. ML
classi?cation models were developed and implemented to assess whether the data
recorded from the sensor system was able to correlate to the correct severity
scores of the three cardinal symptoms of PD. The three classi?cation methods
used in this study were: Simple decision trees,Multi-class support vector
machines (SVMs) and K-nearest neighbours (kNN). In total, 234 data sets (13
separate trials x 6 symptoms x 3 repetitions) were captured.MMG, a measure of
the acoustic or vibrational artefact of muscle movement, as a means of
recording this information. They obtained an average accuracy of 90.9%. They
concluded that ?ne kNN performed best out of the three ML models achieving the
highest accuracy for 5 out of the 6 symptoms.

The
research paper of Galaz, Zoltan, et al. 8 deals with  Parkinson’s disease (PD) severity
estimation according to the Unified Parkinson’s Disease Rating Scale: motor
subscale (UPDRS III), which quantifies the hallmark symptoms of PD, using an acoustic
analysis of speech signals.  They have
used the technique of guided regularized random forest algorithm to select
features with maximum clinical information and performed random forests
regression to estimate PD severity. They obtained their dataset by studying a
total of 100 Spanish native speakers from Colombia were studied , where 50 of
them suffered from PD. They showed that, according to significant correlations
between true UPDRS III scores and scores predicted by the proposed methodology,
information extracted through variety of speech tasks can be used to estimate
the degree of PD severity.
Heng-Tze Cheng et al. 9
used a regression model to calculate the UPDRS score of patients and predict
the severity of Parkinson’s disease, exploiting the harmonic nature of the
voice.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

Results showed that the
epsilon support vector machine with polynomial kernel of degree 3 was found to
be most effective, whose performance was about 5.66 mean absolute error as
measured on a 20-fold cross-validation.

Nilashi, Mehrbakhsh et al.10, proposed
a new hybrid intelligent system for the prediction of PD progression using
noise removal, clustering and prediction methods. Principal Component Analysis
(PCA) and Expectation Maximization (EM) were respectively employed to address
the multi-collinearity problems in the experimental datasets and clustering the
data. Then, Adaptive Neuro-Fuzzy Inference System (ANFIS) and Support Vector
Regression (SVR) was applied for prediction of PD progression. The dataset was
obtained from UCI machine learning repository (Bache and Lichman, 2013). As a
result, average accuracy for Total UPDRS and Motor UPDRS obtained was 47.2%
and  44.3% respectively.
Bhattacharya and Bhatia 11 used data mining tool, Weka, to pre-process the
dataset on which they used Support Vector Machine (SVM) to distinguish people
with PD from the healthy people. They applied LIBSVM to find the best possible
accuracy on different kernel values for the experimental dataset. They measured
the accuracy of models using Receiver Operating Characteristic (ROC) curve
variation.
Chen et al.12 presented a diagnosis PD system by using Fuzzy K-Nearest
Neighbor (FKNN). They compared the results of developed FKNN-based system with
the results of SVM based approaches. They also employed PCA to further improve
the PD diagnosis accuracy. Using a 10-fold cross-validation, the experimental
results demonstrated that the FKNN-based system significantly improve the
classification accuracy (96.07%) and outperforms SVM-based approaches and other
methods in the literature.