MedGEM brings Google's MedGemma and MedASR to offline Android devices for private, multimodal clinical support anywhere.
Secure FedRAG framework for distributed health data search and knowledge exchange
AI-powered hospital management system with MedGemma VLM for automated chest disease analysis, role-based workflows (Nurse/Doctor/Admin), and Wavelet-HAT super-resolution for CXR enhancement.
CardioGuard is an end-to-end health technology platform that collects real-time cardiac data from Holter ECG devices, performs anomaly detection using a multi-layer AI pipeline (CNN + Rule Engine + MedGemma), and provides clinical-grade monitoring for patients and doctors.
Efficient fine-tuning of MedGemma using Unsloth and LoRA for Medical Question Answering. MedGemmaInsight delivers memory-efficient training, faster inference, and accurate healthcare-focused responses through domain-specific adaptation.
A hands-on mini course on multimodal large language models and their clinical applications - from architecture to fine-tuning, inference, and evaluation with MedGemma 1.5.
AEGIS is a multi-agent AI Clinical Assistant built with Flutter and Med-Gemma 4B to help doctors diagnose, retrieve records, and generate reports hands-free in under-resourced clinics.
[ACM AI 2026] Safety-audit framework for vision-language models on brain MRI, measuring calibration, confident errors, hallucination, abstention, and slice-level accuracy.
Production-ready FastAPI service for MedGemma 4B (Q4_K_M GGUF, ~2 GB) using llama-cpp-python, built for efficient local CPU inference with Docker.
MedGemma Impact Challenge (Kaggle Hackathon) : AI-assisted Medical audit system for breast cancer screening compliance.
MedGemma Fine-Tuning for Opentrons Protocol Metadata
MedContext is an agentic AI system for detecting medical misinformation and contextual authenticity using the MedGemma multi-modal model. submitted to the Kaggle MedGemma Impact Challenge.
Medical QA project that fine-tunes a general-purpose LLM using QLoRA on the MedMCQA dataset and benchmarks it against the domain-specific model MedGemma, evaluating in-distribution performance on MedMCQA and out-of-distribution generalization on MedQA.
Automatic brain tumor classification and diagnostic support using magnetic resonance imaging (MRI) scans. It classifies images into four categories: Glioma, Meningioma, Pituitary, or No Tumor.
MedGemma 4b it fine tuned on BBBC006 dataset
Heart — An AI‑powered healthcare assistant built with FastAPI, LangGraph, and RAG pipelines. Provides cardiac risk prediction, medical Q&A, and agentic AI workflows using vector search and local LLMs.
Retrieve biomedical data across decentralized hospital nodes using federated learning and differential privacy to enable secure question answering.
NeuroGuard-NICU: Explainable AI Risk Intelligence for Neonatal Intensive Care
To associate your repository with the medgemma-4b-it topic, visit your repo's landing page and select "manage topics."