Data Science Graduate Student
• Designed and implemented RadiantScriptor, an AI system utilizing natural language processing to autonomously generate radiology reports from medical image labels. This innovation streamlines radiologist workflow and improves efficiency• Conducted extensive cancer survival analysis to uncover patterns across demographics and cancer types. Pinpointed highest and lowest survival rates across different cancers, highlighting disparities by age and race, as well as fluctuations over time.•… Show more • Designed and implemented RadiantScriptor, an AI system utilizing natural language processing to autonomously generate radiology reports from medical image labels. This innovation streamlines radiologist workflow and improves efficiency• Conducted extensive cancer survival analysis to uncover patterns across demographics and cancer types. Pinpointed highest and lowest survival rates across different cancers, highlighting disparities by age and race, as well as fluctuations over time.• Developed a face-based BMI prediction model by fine-tuning the VGG16 Pretrained Image Model, achieving exceptional accuracy. Deployed the model through an interactive web-based interface for real-time BMI prediction using Flask• Collaborated with a team of two to develop a data-driven medical chatbot for the Data Science Institute hackathon, aimed at reducing time demands on medical professionals by 40% and empowering patients with informed health decisions• Employed XGBoost, Light GBM, and Logistic Regression to comprehensively analyze a Diabetes Health Indicators Dataset. Built high-performing models for accurate prediction of diabetes onset. Evaluated model effectiveness using metrics like accuracy, F1 score, and recall, with a focus on recall for optimal patient risk identification• Conducted a large-scale analysis (~100M tweets) using PySpark and Locality Sensitive Hashing (LSH) to assess Twitter's credibility as an information source in the education field. Analyzed the platform's effectiveness in capturing emerging trends and ideas Show less