搜索医学 AI 论文与资源

ARK is a chest radiography foundation model reported in Nature Machine Intelligence for visual representation learning and radiology downstream tasks.

Nature Medicine paper on Med-PaLM 2 and expert-level medical question answering with large language models.

Nature Medicine article describing a generalist biomedical vision-language foundation model evaluated across multiple biomedical tasks.

Prov-GigaPath is a pathology foundation model trained on real-world whole-slide images and released with a Nature paper and project resources.

MedSAM adapts the Segment Anything paradigm to medical image segmentation and reports broad evaluation across imaging modalities.

Med-PaLM Multimodal evaluates a generalist biomedical AI system across medical question answering, imaging, report generation, and multimodal tasks.

一种面向多源异构标注数据的监督训练方法，用于训练单一高性能 AI 基础模型，避免人工标签整合。

NEJM AI 论文，报道基于超过 1000 万条 ECG 记录构建的心电图基础模型。

Nature Communications 论文，提出用于医学文献挖掘的人机协作基础模型。

系统综述与荟萃分析，评估临床医学中人类与大语言模型协作相对于人类单独工作流的表现，覆盖临床推理、文档和解释等任务；研究指出当前证据仍初步且具有情境依赖性，建议后续开展预注册、务实、多中心并嵌入真实工作流的临床研究。

Multimodal clinical data foundation and benchmark introduced at ICML 2025 for clinical foundation model research.

Interactive medical image segmentation benchmark and baseline from CVPR 2025, covering multiple modalities, organs, and target structures.

Google Health AI Developer Foundations open model resources for medical text and medical image understanding, including MedGemma 1.5 resources.

Benchmark for evaluating health AI model safety, helpfulness, and clinical-relevance judgments with physician-reviewed rubrics.

Medical LLM benchmark and leaderboard intended to broaden coverage beyond single medical QA datasets.

Foundation model and toolkit for all-in-one biomedical image parsing across recognition, detection, and segmentation tasks.

Deidentified EHR data for ICU and hospital patients at Beth Israel Deaconess Medical Center, distributed through PhysioNet with credentialed access.

JPG-formatted chest radiographs with labels derived from free-text reports, hosted by PhysioNet.

Large publicly available 12-lead ECG waveform dataset with diagnostic labels, hosted on PhysioNet.

Stanford chest radiograph dataset for automated chest X-ray interpretation and uncertainty-aware label evaluation.

NIH Clinical Center chest X-ray dataset released for computer-aided detection and radiology machine learning research.

Legacy multi-task biomedical image segmentation benchmark retained as a reference; newer segmentation benchmarks are listed above it.

Low-dose CT imaging collection from the National Lung Screening Trial, distributed by The Cancer Imaging Archive.

MICCAI 2026 challenge for pathologist reasoning-guided pathology report generation, hosted on Grand Challenge.

PhysioNet Challenge 2026 on detecting cognitive impairment from polysomnography and related clinical signals.

HL7 healthcare AI challenge focused on AI applications around health data interoperability and standards-based workflows.

MICCAI Student Board educational challenge for tutorial-style submissions around medical image computing education.

RSNA announced a 2026 AI challenge focused on accelerating knee MRI examinations and reconstruction quality.

Medical Imaging and Computer-Aided Diagnosis 2026 call for full papers, posters, and oral presentations.

Stanford Medicine RAISE Health symposium on responsible AI for safe and equitable health care.

DeepLearning.AI specialization on diagnosis, prognosis, and treatment using medical AI workflows.

Stanford Center for Artificial Intelligence in Medicine and Imaging course page covering AIMI short courses and programs.

RSNA certificate program for radiology professionals applying AI in medical imaging practice.

MIT course materials for machine learning methods in healthcare and clinical data.

Harvard Medical School HMX Pro online course on applying artificial intelligence in medicine.

Stanford AIMI Grand Rounds seminar series on artificial intelligence in medicine and imaging.