{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreihgm3utf2lbxmwzgaj5eqloq3q7icddwwhoynaiazrpupl5ciwp4y",
"uri": "at://did:plc:2gbt2dlwaqovtnmxkat3tyke/app.bsky.feed.post/3mgas3niuma72"
},
"coverImage": {
"$type": "blob",
"ref": {
"$link": "bafkreic6bed4zg7m4hjtxm7r5tuyfehoyitooc2m3zarzdwvfhrtfpfite"
},
"mimeType": "image/png",
"size": 162062
},
"path": "/articles/s41586-026-10181-8",
"publishedAt": "2026-03-04T17:18:46.149Z",
"site": "https://www.nature.com",
"tags": [
"doi:10.1038/s41586-026-10181-8"
],
"textContent": "Nature, Published online: 04 March 2026; doi:10.1038/s41586-026-10181-8\n\nMerlin, a vision–language foundation model trained on a large dataset of paired CT scans, patient record data and radiology reports, demonstrates strong performance across model architectures, diagnostic and prognostic tasks, and external sites.",
"title": "Merlin: a computed tomography vision–language foundation model and dataset",
"updatedAt": "2026-03-04T00:00:00.000Z"
}