Parkinson’ s disease detection using vocal biomarkers.pptx

This presentation is about Parkinson's disease detection using vocal biomarkers

1. Parkinson’ s disease detection using vocal biomarkers Presented by: Bhanu Pratap Singh

2. Introduction What is Parkinson’s disease? A progressive neurodegenerative disorder characterized by the loss of dopamine-producing neurons in the brain, leading to impaired movement and other symptoms. Common Symptoms Tremors, Muscle rigidity, Slowness of movement (bradykinesia) Changes in speech and facial expression Impact and Prevalence Affects over 10 million people worldwide Second most common neurodegenerative disorder after Alzheimer's Importance of Early Diagnosis Early detection is critical to slow disease progression, improve treatment outcomes, and enhance the quality of life for patients.

3. Role of Vocal Biomarkers in Parkinson's Disease Detection What are Vocal Biomarkers? Changes in voice characteristics (such as pitch, tone, loudness, tremor, and speech rate) that may indicate neurological changes. Why Use Vocal Biomarkers for PD? Early Indicators : Voice changes can occur even in the early stages of Parkinson's, potentially before more obvious symptoms. Non-Invasive : Voice analysis does not require complex medical procedures or expensive equipment. Affordable and Accessible : Can be easily integrated into telehealth applications, making it suitable for large-scale screening. Relevant Research Findings Little et. El. (2009) achieved a classification performance of 91.4%, using a kernel SVM. Sadek , Ramzi M., et al. "Parkinson’s disease prediction using artificial neural network." (2019). Image source: samoonmd.com

4. Frameworks and tools Scikit-learn : For implementing SVM, Random Forest, and feature selection methods. Pandas : For data manipulation and preprocessing tasks. NumPy : For numerical computations and array manipulations. Matplotlib/Seaborn : For data visualization and understanding data distributions. TensorFlow: For building and training neural network models. Image source: interviewbit.com

5. Proposed approach and methodology 1. Dataset Using the Parkinson's dataset by Little, M. (2007), obtained from the UCI Machine Learning Repository. The dataset includes voice recordings from individuals, with features related to voice frequency and variation The status column differentiates between healthy individuals (0) and those with PD (1). 2. Data Preprocessing Normalization : Scale the features to a common range. Feature Extraction : Identify and extract the relevant features from the dataset. .

6. 3. Model Selection Choose an appropriate machine learning algorithms, such as: Linear classification Support Vector Machines (SVM): Random Forests Neural Networks 4. Training and testing Divide the dataset into training and testing sets (e.g. 70% training, 30% testing) to evaluate model performance Image source: sciencedirect.com

7. 5. Performance Metrics Confusion Matrix: Provides detailed insight into model’s performace . Key Metrics Accuracy The proportion of correct predictions made by the model. (TP+TN)/(TP+TN+FP+FN) Provides a general sense of model performance. Precision The proportion of true positive predictions among all positive predictions. TP/(TP +FP) Indicates the quality of positive predictions. Confusion matrix ( Image Source: ibm.com )

8. Recall (Sensitivity) The proportion of true positives among all actual positives. TP/(TP+FN) Measures the model's ability to identify actual positive cases (PD). F1 Score The harmonic mean of precision and recall. 2×( Precision×Recall )/( Precision+Recall ) Balances precision and recall, useful for imbalanced datasets.

9. Thank you!