Koushika Muddana

Project Objective: Developed a machine learning-based system for early prediction of liver disease using algorithms like Random Forest and ensemble methods.
Key Achievement: Identified Random Forest as the most accurate model, achieving high performance across accuracy, precision, recall, and F1 score metrics.
System Design: Implemented data preprocessing, feature selection, and model evaluation; integrated Flask for user interaction.
Future Scope: Proposed deep learning models and ROPE analysis for enhanced feature evaluation and prediction accuracy.

Project Overview: Implemented a Twitter Sentiment Analysis system using Naïve Bayes Classifier to automatically classify customer sentiment, optimizing social media customer feedback analysis.

Key Techniques: Conducted text preprocessing (tokenization, punctuation removal, stop word elimination) and count vectorization to prepare data for classification, leveraging Scikit-Learn.

Performance Evaluation: Assessed model accuracy using confusion matrices and fine-tuned the Naïve Bayes Classifier for improved sentiment classification.

Visualization and Tools: Applied Python libraries for data visualization, including word clouds, and built an automated pipeline for efficient text data cleaning and tokenization.

Project Overview: Designed a "Critical Patient Management System" (CPMS) using machine learning models and IBM Cloud (PaaS) to predict patient conditions in real-time, enhancing critical healthcare management.

Technological Implementation: Developed and deployed models such as Random Forest, Logistic Regression, and Gradient Boosting, utilizing IBM Watson Studio for auto-deployment and cloud-based data management.

Key Features: Integrated an SMS notification system to alert doctors and caregivers of critical health changes, with a mobile application for real-time monitoring of patients' vitals.

Results Achieved: Achieved model accuracies ranging from 80% to 92% with ensemble methods, successfully implemented real-world testing with a 90% accuracy rate in patient condition prediction and notifications.