Microsoft Health Futures
ダウンロード
Protein Language Model Subnetworks
6月 2025
Protein language models (PLMs) pretrained via a masked language modeling objective have proven effective across a range of structure-related tasks, including high-resolution structure prediction. However, it remains unclear to what extent these models factorize protein structural categories among their learned…
Immunomics – JL-GloVe
3月 2025
We employ GloVe and random projection theory to infer immunologically meaningful T-cell receptor embeddings from adaptive immune repertoires. This repository contains the Pytorch code to replicate experiments in our paper “Scalable Universal T-Cell Receptor Embeddings from Adaptive Immune Repertoires” accepted…
RAD-DINO model
11月 2024
RAD-DINO is a vision transformer model trained to encode chest X-rays using the self-supervised learning method DINOv2. RAD-DINO is described in detail in RAD-DINO: Exploring Scalable Medical Image Encoders Beyond Text Supervision (F. Pérez-García, H. Sharma, S. Bond-Taylor, et al., 2024).
MAIRA-2 model
11月 2024
MAIRA-2 is a multimodal transformer designed for the generation of grounded or non-grounded radiology reports from chest X-rays. It is described in more detail in MAIRA-2: Grounded Radiology Report Generation (S. Bannur, K. Bouzid et al., 2024). MAIRA-2 has been built…
RadFact: An LLM-based Evaluation Metric for AI-generated Radiology Reporting
11月 2024
RadFact is a framework for the evaluation of model-generated radiology reports given a ground-truth report, with or without grounding. Leveraging the logical inference capabilities of large language models, RadFact is not a single number but a suite of metrics, capturing aspects of precision…
PadChest-GR dataset
11月 2024
PadChest-GR is a manually annotated, bilingual chest X-ray dataset designed to train and evaluate models for grounded radiology report generation. It includes bounding boxes and comprehensive annotations of all clinically relevant findings.
MICON
10月 2024
This is the repository for paper “Causal integration of chemical structures in self-supervised learning improves representations of microscopy images for morphological profiling”.
ProtNote: a multimodal method for protein-function annotation
10月 2024
ProtNote is a multimodal deep learning model that leverages free-form text to enable both supervised and zero-shot protein function prediction.
RadFact
8月 2024
RadFact is a framework for the evaluation of model-generated radiology reports given a ground-truth report, with or without grounding. Leveraging the logical inference capabilities of large language models, RadFact is not a single number but a suite of metrics, capturing…