A novel medical large language model family with 13/70B parameters, which have SOTA performances on various medical tasks

Cross-type Biomedical Named Entity Recognition with Deep Multi-task Learning (Bioinformatics'19)

Bioformer: an efficient BERT model for biomedical text mining

[EMNLP 2024] This is the code for our paper "BMRetriever: Tuning Large Language Models as Better Biomedical Text Retrievers".

This repository contains the code used for distillation and fine-tuning of compact biomedical transformers that have been introduced in the paper "On The Effectiveness of Compact Biomedical Transformers"

Systematic evaluation of hallucination risks in Large Language Models (GPT-4, Claude 3, Gemini Pro) for clinical proteomics and mass spectrometry interpretation. Production-ready detection framework with comprehensive benchmarks.

Patient Family History extraction tool

Graph-based RAG system for biomedical nutrigenetic knowledge discovery. Enables natural language queries on gene-nutrient interactions, supports personalized nutrition counseling, and runs 100% locally with Ollama LLMs and SBERT embeddings.

BERT-for-BioNLP-OST2019-AGAC-Task2

AGAC-BioNL-OST2009-Task1 BERT+CRF

BioCreative VIII track 2

Implements relation extraction for biomedical texts using Hard Negative Mining to improve accuracy in identifying complex entity relationships. Includes code for data processing, training, and evaluation with BioC-format datasets.

Cancer-Alterome is a comprehensive and curated dataset that focuses on the investigation of regulatory events caused by gene alteration in the context of cancer.

MSR Cambridge Internship Summer 2023

RAG pipeline for medical question-answering. Fuses lexical and dense retrieval (MedCPT, Contriever, Specter + FAISS) with OpenAI, Gemini, and HuggingFace LLMs. Supports iterative multi-round reasoning, strict typing, structured observability, and a clean layered architecture

Core LLM for M.A.R.S. (Model Assisted Review System). Utilizes fine-tuned Llama 3.2 3B to automate biomedical SLR screening with 92.2% accuracy.

BioGemma — Google Gemma 3 1B fine-tuned on medical/biomedical corpus for clinical NLP tasks

MedQA-NLI is a comprehensive medical reasoning dataset comprising 42,889 instances designed for training and evaluating models on natural language inference (NLI) tasks in biomedical domains.

Clinical trial document intelligence pipelines using medallion architecture. Classification (87 categories) + NER (8 entity types) on Databricks.