Machine Learning in Diagnosis and Treatment Essay Example

Editorial." BioMedical Engineering OnLine. Accessed May 16, 2016. http://www.ncbi.nlm.nih.gov/pmc/ 2: "Health Information." NIDCD. Accessed May 15, 2016. https://www.nidcd.nih.gov/.

Machine learning algorithms usually involve three stages in the solution of VFP: Feature Extraction, Feature Reduction, and ClassificationfootnoteRef:3. In the proposed method, discrete wavelength transform is used to extract features, Principal Component Analysis (PCA) and Kernel Principal Component Analysis (KCPA) for feature reduction. PCA involves conversion from a space of high dimensions to a more reduced size; it is an ideal method since within it, there is a decreased redundancy of information is; KPCA is a non-linear version of PCA developed by use of the kernel method. Artificial Neural Network (ANN) is applicable in the classification of speech sounds and detection of articulation disorders.

• Majidnezhad, Vahid, and Igor Kheidorov. "An ANN-based Method for Detecting Vocal Fold Pathology." International Journal of Computer Applications IJCA 62, no. 7 (2013): 1-4. doi:10.5120/10089-4722.

Feature extraction means finding suitable parameters that help to classify between the healthy and abnormal patientsfootnoteRef:4. VFP results in disruption of the ordinary flow of voice; we take Discrete Wavelength Transform (DWT) for time-frequency analysis of speech signals. In the Laryngeal electromyography diagnosis, a fine needle is inserted into the muscles of the Adam's apple footnoteRef:5. The patient wears a grounding electrode and these, in turn, are connected to a computer, through which the specialist observes the pattern of the wavelengths. DWT is an enactment of the wavelet transform by use of a distinct set of the wavelet scales and translations obeying some well-defined procedures; this transform decomposes the signal into a conjointly orthogonal set of wavelets. A comparison is carried out between wavelets with different support-sizes. Notably, the proposed

approach, implemented with modest computer requirements, results in an adequate voice box physiology classifier for identification of nodules in vocal folds.

• Khosrow-Pour, Mehdi. Assistive Technologies: Concepts, Methodologies, Tools, and Applications. Hershey, PA: Information Science Reference, 2014. 5: "Diseases and Conditions." -. Accessed May 16, 2016. http://www.mayoclinic.org/diseases-conditions/vocal-cord-paralysis/basics/definition/con-20026357

In the PCA method, the dimension of the feature vector is decreased. This process seeks a mapping to find the best way to represent the distribution of datafootnoteRef:6. It, therefore, uses a signal representation criterion to perform dimension reduction while conserving much of the unpredictability or variance in the high-dimensional space as possible. PCA involves the calculation of the eigenvalues breakdown of a data covariance matrix or singular value decomposition of a data matrixfootnoteRef:7.

• Majidnezhad, Vahid, and Igor Kheidorov. "An ANN-based Method for Detecting Vocal Fold Pathology." International Journal of Computer Applications IJCA 62, no. 7 (2013): 1-4. doi:10.5120/10089-4722. 7: "Principal Component Analysis." Principal Component Analysis. Accessed May 16, 2016. http://www.fon.hum.uva.nl/praat/manual/Principal_component_analysis.html

Usually after mean centering the data for each attribute. In the last step, classification, the speech signal is classified into two classes: pathological or healthy by ANN (Artificial Neural Network) or SVM (Support Vector Machine) methods. Just like every other computing system, ANN is made up of some interconnected processing elements, which processes information by its dynamic state response to external inputs. A set of input neurons whose activation is the wavelength of the voice for speech recognition defines a neural network. After weighting and transformation, the activations of these neurons are then passed on to other neurons.

This process continues until finally, the output neuron that determines the normal/pathological speech is activated. Since ANNs are pro's at distinguishing patterns, they

are trained to generate an output when something unusual occurs that misfits the wavelength outline. Under SVM, 23 wavelets types in two experts SVM machines conducts tests, the first using "one against all" strategy to differentiate normal and pathological voices and the second using "one against one" to classify pathologies: Hence, they accurately distinguish the healthy from the pathological speech.

Bibliography

• "ANN-based Method for Detecting Vocal Fold Pathology." Accessed May 15, 2016. http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.303.7065.
• "Health Information." NIDCD. Accessed May 15, 2016. https://www.nidcd.nih.gov/.
• Afsah, Omaymae. "Approach to Diagnosis of The Vocal Fold Immobility: A Literature Review." Egypt J Otolaryngol The Egyptian Journal of Otolaryngology 31, no. 2 (2015): 78. doi:10.4103/1012-5574.156088.
• Alvin. "BioMedical Engineering OnLine: An Editorial." BioMedical Engineering OnLine. Accessed May 16, 2016. http://www.ncbi.nlm.nih.gov/pmc/
• Majidnezhad, Vahid, and Igor Kheidorov. "An ANN-based Method for Detecting Vocal Fold Pathology." International Journal of Computer Applications IJCA 62, no. 7 (2013): 1-4. doi:10.5120/10089-4722.
• Simpson, C. Blake, and Esther J. Cheung. "Evaluation of Vocal Fold Paralysis." Vocal Fold Paralysis: 55-62. Doi: 10.s1007/3-540-32504-2_4.