Machine Learning Approaches to Dysarthria Severity Analysis

About Internship

Assistant Professor, Department of Mechanical Engineering, from Indian Institute of Technology, Patna (IIT Patna), is accepting interns interested in the field of Computer Science Engineering, in the research area of Artificial intelligence and Machine learning.

The project aims to develop robust computational methods for assessing the severity of dysarthria from speech signals, thereby supporting objective diagnosis and rehabilitation monitoring. Candidates will be responsible for handling and preprocessing dysarthric speech datasets, which includes tasks such as cleaning, noise removal, segmentation, and annotation of severity levels to ensure high-quality input for analysis. A major responsibility will involve extracting relevant acoustic and prosodic features such as MFCCs, jitter, shimmer, formants, and energy-based measures, alongside exploring temporal-spectral representations like spectrograms and mel-spectrograms. They will implement and evaluate fundamental machine learning algorithms including SVM, Random Forest, Decision Trees, k-NN, and Logistic Regression to establish baseline classification models, with scope to explore advanced deep learning methods depending on project requirements. Candidates are expected to optimize models, tune hyperparameters, and assess system performance using appropriate evaluation metrics such as accuracy, F1-score, unweighted average recall, and correlation coefficients, ensuring statistically valid outcomes through cross-validation. Beyond technical implementation, they will maintain structured documentation of datasets, code pipelines, and experiment logs, while preparing regular progress reports and presentations to facilitate collaboration. Active participation in literature review, result interpretation, and drafting research manuscripts or technical deliverables is also expected. This project provides an opportunity for candidates to enhance their expertise in biomedical speech signal processing, feature engineering, and applied machine learning while contributing to clinically meaningful research that can improve diagnostic tools for individuals with dysarthria.