Medical AI

Med-MMFL provides the first comprehensive multimodal federated learning benchmark to enhance medical AI research while maintaining data privacy.

Self-MedRAG enhances medical question answering reliability by integrating hybrid retrieval and iterative self-reflection.

Develop an advanced benchmark for evaluating and improving medical AI agents' clinical reasoning using real-world EHR data.

Enhanced Pix2Pix framework optimizes MRI imaging with few-shot learning for faster, cost-effective medical diagnostics.

DSA-SRGS enhances resolution in dynamic 4D angiography models, improving cerebrovascular diagnosis precision.

Develop an AI-powered tool to improve nurse training in endotracheal suctioning through video-based activity recognition using explainable AI.

AI-powered platform aiding biomedical technicians in LMICs with real-time medical equipment diagnostics and repair.

Zero-shot body region detection system for CT and MR imaging enhances medical analysis by eliminating reliance on DICOM metadata.

Attention-guided attribution for multimodal clinical summarization improves transparency and accuracy without retraining models.

AgentsEval enhances the clinical reliability of AI-generated medical imaging reports by using multi-agent stream reasoning for evaluation.

Develop AI models for detecting surgical tools using large vision-language models to enhance surgical guidance.

A benchmark study evaluating the performance of classical ML versus transformer-based models in medical classification tasks.

Atlas 2 offers state-of-the-art pathology vision models designed for clinical deployment with enhanced performance and efficiency.

Health-SCORE makes rubric development for evaluating Health-LLMs more scalable and cost-effective.

MARBLE offers a scalable, efficient tool for multi-scale whole-slide image analysis with significant accuracy improvements.

MetaDrug is a meta-learning framework for personalized medication recommendation addressing the cold-start problem in new patients using EHR data.

Develop a high-accuracy EEG decoding tool for neurological applications leveraging efficient model architectures.

A unified 3D medical vision-language model for advanced multimodal reasoning and precise 3D segmentation.

AI-based diagnostic tool for accurate pediatric pneumonia detection using explainable deep learning.

Develop a clinically-grounded AI tool for accurate heart rate variability interpretation in biomedical engineering.

LeMoF provides a novel multimodal fusion framework enhancing clinical predictions from heterogeneous datasets like EHRs and biosignals.

A workflow to optimize downsampling of nEMG signals for efficient neuromuscular disease detection.

MedRedFlag enhances LLM safety in medical communication by curating a dataset for better redirection of misconceptions in patient queries.

Develop a tool for learning temporal embeddings from electronic health records to improve chronic kidney disease patient outcomes.

Develop Fair-Eye Net, an AI system for equitable glaucoma screening and follow-up using multimodal data integration.

EHR-RAG enhances clinical prediction from long-horizon EHRs using advanced retrieval-augmented generation techniques.

MHub.ai is a standardized platform for deploying and reproducibly benchmarking AI models in medical imaging.

AI model for resilient multimodal survival prediction in NSCLC that handles missing modalities effectively.

A pipeline enabling path-specific causal fairness for machine learning models in healthcare to address biases.

A Planner-Auditor framework enhances clinical discharge planning reliability using FHIR data and self-improvement loops.

ReaMIL enhances multiple instance learning in histopathology by efficiently selecting evidence while maintaining accuracy.

Develop AI-assisted drug designs to promote degradation of toxic proteins in Alzheimer's using molecular glues.

Develop an open-source framework for aligning language models with clinical trial embeddings for improved interpretability and generative use cases.

Medical SAM3 delivers a universal, prompt-driven segmentation model for medical imaging, solving domain shift challenges.

DeepMed enhances medical reasoning models by grounding in verifiable evidence and improving accuracy on medical benchmarks.

AI-enhanced tool to augment clinician response drafting in patient portals by aligning LLM outputs with clinician preferences.

Advanced image segmentation tool preserving topological attributes and width, ideal for medical imaging applications.

AI-powered breast cancer classification using synthetic data from advanced imaging.

MetaboNet offers a standardized, consolidated dataset for type 1 diabetes management, poised to become the benchmark for AI-driven diabetes intervention technologies.

Develop a dermatology visual question answering tool utilizing the DermaBench dataset for enhanced clinical decision support.

APOLO boosts emotion analysis in mental health diagnostics with a multi-agent prompt optimization framework.

Develop a CT triage system with organ-aware attention to improve radiology workflow efficiency.

Develop CLEAR-Mamba to enhance ophthalmic angiography classification's accuracy and reliability using hypernetworks and evidential uncertainty learning.

A novel graph-based method for accurate brain-tumor localization in multi-modal MRI using a decoder-free architecture.

Develop a noise-robust medical image segmentation tool using an abstention framework to improve accuracy in noisy datasets.

AI-powered tool for predicting surgical outcomes in chronic rhinosinusitis patients using a blend of ML and GenAI.

GeoDynamics offers a novel geometric state-space neural network customizing brain dynamics analysis on Riemannian manifolds for early disease detection.

Develop on-device ECG dialogue agents capable of multi-turn interactions to improve cardiac diagnostics.

Automated rubric generation for evaluating and refining medical dialogue systems.

A federated learning platform for privacy-preserving brain tumor localization across healthcare institutions.