Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

A Study on Telemonitoring of Parkinson’s Dise

No description
by

novel geek

on 30 April 2014

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of A Study on Telemonitoring of Parkinson’s Dise

Degenerative Disorder in central nervous system.
The brains cells that make  ”Dopamine” slowly die.
Parkinson's disease most often develops after age 50.
1 million of people in US are living with Parkinson.

Parkinsons disease
A Study on Telemonitoring of Parkinson’s Disease Progression using Data Mining Techniques.

Symptoms
Rating Scales
Progression is measured by different rating scales.
Hoehn and Yahr scale, UPDRS, Schawb and England ADL scale
UPDRS is the most prominent
UPDRS
Results
voice signal attributes for classifcation.
Expected results
Constraints
Time
Data set
Necessary attributes
Classification methods used
J48
NNge
PART
Ridor
BFTree
Implementation
Sireesha Kakumanu Lalitha Pragna Nasina Bryce Cooper
Univeristy of Nebraska Lincoln

22 attributes
4 independent
16 voice dependent measures
UPDRS dependency
Co-relation among different attributes
Association rules
age='(75.2-80.1]' sex='(-inf-0.1]' 52 ==>motor_UPDRS='(22.2744-25.72172]'
age='(50.7-55.6]' sex='(-inf-0.1]' 45 ==> motor_UPDRS='(15.37976-18.82708]'
age='(70.3-75.2]' sex='(0.9-inf)' 67 ==> motor_UPDRS='(15.37976-18.82708]'
age='(55.6-60.5]' sex='(0.9-inf)' 112 ==> motor_UPDRS='(11.93244-15.37976]

Observed results
Apriori:
Numeric attributes
Less run time
Lot of rules generated
Different minimal support and confidence values
Algorithm
Age and test time influence the UPDRS scores
Speech attributes are used for the calculation of UPDRS
age (40-50) ==> UPDRS
jitter , shimmer ==> UPDRS

Expected results

Weekly tests at home using Intel At Home Testing Device (AHTD).
42 Subjects
sustain “ahh” sound as long and steady as possible.
Total of 5875 signals.( Instances)
22 Attributes (16 biomedical voice measures).

Data set
Objectives
Preprocessing
Attributes
- - Speech disorder
- Tremor of hands, arms, legs
- Slowness of movements
- Stiffness of limbs
- Postural instability
U
nified
P
arkinson's
D
isease
R
ating
S
cale
Performed clincally and has drawbacks
The attributes in the dataset Motor-UPDRS and total-UPDRS were weighed as baseline, three months, six months.

Classification
Classification based on the Total-UPDRS scores
Classification
Stage 1 : Normal : 0-34
Stage 2 : Mild : 35-69
Stage 3 : Slight : 70-104
Stage 4 : Moderate : 105-139
Stage 5 : Severe : 140 - above
Conclusions
Classification trees are complicated
NNge is the optimal method
Classify people based on voice measures
Subjects fall into stages 1 and 2

Association Rules
Association between different attributes
Prediction
Predict total UPDRS scores for further cases
Conclusions
Depends on Correlation Coefficient.
Model with highest correlation coefficient is effective model.
Predicting Total_UPDRS score based on the attributes.
UPDRS prediction
Prediction model for UPDRS
Telemonitoring Data Attributes
Conclusions
Co-relation Coefficient
Hierarchical breakdown of the data.
builds a model where the class is known.
use that model to classify new data where the class is unknown.
Joins an inducer algorithm ( in java).
Implemented cross validation using 10 folds as test options

Prediction using decision table model
Total UPDRS
vs
Predicted Total UPDRS
Prediction of UPDRS helps in identifying the severity of disease.
Helps in diagnosis.
Decision table produces accurate results.
This model can be also further in predicting the scores.
Building and implanting the model was challenging.

Conclusion
Questions?
References
Data Mining: Practical Machine Learning Tools and Techniques
http://www.cs.waikato.ac.nz/ml/weka/book.html
Linguistic Instituite Universality & variability.
http://languagelog.ldc.upenn.edu/myl/lsa2013/TheoriesAndData.html
Modeling and Predicting with the Decision Table Classifier and Visualizer http://techpubs.sgi.com/library/dynaweb_docs/0640/SGI_EndUser/books/MineSet_UGNT/sgi_html/ch09.html
National Institute of Neurological Disorders and Stroke http://www.ninds.nih.gov/disorders/parkinsons_disease/parkinsons_disease.htm
Data mining using voice measurements increased the chance of early diagnosis.
Data mining approaches helped in obtaining useful relations.
But applying approaches in this data set was challenging and interesting.
As further work , the time series approach can be implemented with more than available data.
Jitter:DDP='(-inf-0.016566]' 149 Jitter:RAP='(-inf-0.005519]' 149 ==> total_UPDRS='(21.3976-26.1968]'
Jitter:RAP='(-inf-0.005519]' 149 Jitter:DDP='(-inf-0.016566]' 149 ==> total_UPDRS='(21.3976-26.1968]'
Jitter:DDP='(-inf-0.016566]' 140 Jitter:RAP='(-inf-0.005519]' 140 ==> total_UPDRS='(26.1968-30.996]'
Jitter:RAP='(-inf-0.005519]' 140 Jitter:DDP='(-inf-0.016566]' 140 ==> total_UPDRS='(26.1968-30.996]'

Age and Gender play a vital role in the progression of the disease.
Speech attributes can be used for determination of UPDRS scores.
Attribute values pose a constraint while detrmining the rules.
Telemonitoring UPDRS
Full transcript