This application is a Predictive Model developed for finding severity of hearing Loss in children. It uses a Lasso regression model trained on a dataset of hearing health indicators to predict the risk of hearing loss. The model achieved an accuracy of 87 percent during testing. The tool takes various inputs such as age, gender, and specific health indicators, and provides a risk assessment and recommendation based on the model's prediction.
The project was led by Principal Investigator Dr. Kanipakam Sunitha, Assistant Professor, Dept. of Law, SPMVV, Tirupati, and Co-Principal Investigator Dr. N.V. Muthu Lakshmi, Assistant Professor, Dept. of Computer Science, SPMVV, Tirupati. The project received financial support from the DST-CURIE-AI Center, SPMVV, Tirupati.
The data set was collected from the children who are suffering from hearing impaired and taking training at special school. Children are from different areas of Andhra Pradesh and Telangana.
The initial dataset used for the project contained various features such as gender, age, parental education, hearing loss percentage, and several health indicators. The data was cleaned, unnecessary columns were removed, and categorical data was converted to numeric for model training. Multiple regression techniques were tested, including linear regression, decision trees, random forest, Lasso regression, and Support Vector regression. Lasso regression performed the best, achieving a maximum accuracy of 87%.
The tool is designed to be user-friendly, with a simple interface for inputting information and receiving the risk assessment and recommendations. It serves as a valuable resource for assessing hearing health risks and taking necessary precautions or treatments. However, it's important to consult with a healthcare professional for a comprehensive evaluation. This tool is intended to supplement, not replace, professional medical advice.