• In this project, we have analyzed speech signals for disease prediction. The analysis is done for Parkinson’s Disease (PD) and uses various machine learning based classification methods.
• Research Oriented Project;
• Used multiple feature selection algorithms, linear and non-linear classifiers, various Ensembles, AdaBoost and XGBoost to get the best possible results for Parkinsons Disease Prediction;
Wearable medical sensing and actuating devices with wireless capabilities have become the cornerstone of many revolutionary digital health applications. Devices like these are equipped with motion and audio sensors. These sensors therefore can acquire signal data like heart rate, frequency of speech, breath rate etc. Technical advancements in this field has led to more and more data in the above form.
Since more data are available through this process,the preliminary diagnosis of the disease can be possible; hence diseases can be preventable.
• Parkinson's Disease belongs to one of the categories of neuro-degenerative disease which directly as well as indirectly affects the brain cells that will affect the movement, speech and other cognitive parts.
• As the disease progresses more than 90% of the patients have speech disorders. The symptoms related to the vocal impairment of Parkinson’s disease patients is called dysphonia.
• As a result, medical professionals rely on indicators related to dysphonia to assess the PD patients. These measures/indicators related to dysphonia are important and reliable methods to assess the voice related problem and monitor it at different stage.
• With respect to the past research it is found that artificial intelligence and machine learning techniques have potential for the classification and it was also found that the classification system helps to improve the accuracy and the reliability of the diagnosis.
