biobert

NLP project using Transformer models (BERT and BioBert with Weighted Loss) to automatically classify sentences describing Adverse Drug Events (ADEs) in healthcare text.

Production clinical NLP pipeline — BERT/BioBERT text classification, NER, and sentiment analysis on medical records. 93% classification accuracy. HuggingFace · spaCy · FastAPI · MLflow · Docker.

Human-in-the-loop biomedical NER system using BioBERT to identify disease-related healthcare information from medical text.

Comprehensive Medical RAG system with hybrid retrieval and multi-vector database support.

Doctor In Hand is an AI-Powered disease prediction system that uses a Fine-Tuned BioBERT model to predict the diseases from user symptoms and recommend the appropriate medical specialist.

Testing whether a general LLM can handle clinical NLP with no medical training. GPT-4.1-nano (zero-shot) vs BERT and BioBERT on three tasks: decoding abbreviations, tracking medication status, extracting conditions. The generalist won on precision, without fine-tuning.

utilized biobert for disease identification from user text. tested other models too. dataset used:

🧠 Multimodal Agentic RAG system for medical imaging — UNet-ResNet34 segmentation + BioBERT RAG + LangGraph reasoning → structured differential diagnosis

Team project: AI medical assistant that combines RAG over live PubMed research, BioBERT embeddings and Llama/DeepSeek models with patient data to support clinical decisions.

MediOrbit / CarePath AI — FastAPI + React medical assistant with Claude-powered chat, BioBERT medical NER, X-ray analysis, and prescription OCR over a 23-hospital Tamil Nadu database.

Classifies 32 cancer types from pathology reports: 94.5% held-out accuracy with leakage-safe evaluation.

A comparative study of SVM, Random Forest, and BioBERT models for enhancing medical meta-analysis through accurate literature classification.

Ayush _Backend – A backend service built using [Django / Flask / Node.js / whichever you used] (specify) to power the Ayush application. It features RESTful APIs, authentication, database integration, and scalable architecture ready for production deployment.

RAG pipeline that converts complex medical PDFs into plain-language explanations using BioBERT embeddings, FAISS, and GPT-4o-mini with safety guardrails.

BioBERT-based text classifier to predict PRETOX-related text (PRETOX_REL vs NO_PRETOX_REL).

Multimodal adverse drug reaction prediction using BioBERT + XGBoost fusion with SHAP explainability — built on FDA FAERS data

BioBert Enhanced Patient Treatment Similarity Analysis with Sentence Transformers

Multi-modal cancer survival prediction: Cox PH, XGBoost-AFT, DeepSurv, BioBERT clinical notes, SHAP, and a thread-built visual system. Models beat the structured-data oracle on a cohort with planted ground truth.

Fine-tuning pre-trained BERT architectures from Hugging face