Individual Projects
Identification of Key Taxonomic and Metabolic Players in the Gut Metagenome of T2D Patients
NGS
Metagenomics
Bash Scripting
Bioinformatics
Statistical Analysis
- Identified exclusive CAZyme proteins and Taxa associated with T2D patients through comprehensive metagenomic analysis
- Developed custom bioinformatics pipelines for large-scale metagenomic data processing
- Employed advanced statistical approaches to identify significant microbial biomarkers
- Implemented machine learning models to predict disease states based on microbial profiles
- Published findings in peer-reviewed journal with impact factor > 5.0
Collaborative Projects
Machine Learning Identifies Microbial Biomarkers of Environmental Antibiotic Exposure
Machine Learning
R Programming
Python
Antimicrobial Resistance
Biostatistics
- Conducted metagenomics analysis to identify environmental antimicrobial resistance genes across diverse ecosystems
- Developed predictive models using Random Forest and XGBoost algorithms with 87% classification accuracy
- Implemented analysis pipelines in R and Python for high-throughput data processing and visualization
Gut Microbial Strains – Non-Small Cell Lung Cancer Treatment: Decoding the Connection
Microbiome
Immunotherapy
Statistical Modeling
Clinical Research
Data Visualization
- Analyzed gut microbiome data from 150+ NSCLC patients undergoing immunotherapy treatment
- Developed machine learning models to associate microbial variants with treatment responses (AUC = 0.82)
- Identified 12 potential microbial biomarkers for treatment efficacy prediction
- Published findings in peer-reviewed journal with impact factor > 5.0
Radiology Report Generation using Multimodal AI and LLaMA
LLaMA-2
ResNet-50
FAISS
Transformer
Medical AI
- Created a RAG-based pipeline integrating ResNet-50, Transformer, and LLaMA-2 architectures for medical imaging
- Utilized FAISS vector search for clinical case retrieval to enhance contextual relevance of generated reports
- Achieved 92% accuracy in automated radiology report generation validated by board-certified radiologists
- Reduced report generation time from 15 minutes to under 2 minutes per case
AI Mental Health Chatbot
BERT
OpenCV
CNN
Multimodal AI
NLP
- Built and deployed an AI chatbot with real-time emotion detection capabilities using facial recognition and voice analysis
- Integrated text, speech, and facial recognition for comprehensive mental health assessment and triage
- Implemented crisis detection algorithms with 89% accuracy in identifying high-risk cases