[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-method-teaches-ai-to-respect-medical-code-hierarchies":10,"sections":41},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":30,"tags":31,"sources":36,"feedback":40,"feedback_at":22,"cost_usd":40,"total_tokens":40},8297,"new-method-teaches-ai-to-respect-medical-code-hierarchies","New Method Teaches AI to Respect Medical Code Hierarchies","An unreviewed arXiv preprint claims a new embedding technique beats rivals at clinical prediction tasks using ICD and ATC code structures.","Researchers have a new way to make medical AI models understand that diseases and drugs come in family trees, not flat lists.\n\nThe method, called HCOE (Hyperbolic Clinical Ontology Embeddings), takes existing BioBERT language model outputs and remaps them into a curved geometric space called a Poincare ball. That space is naturally suited to representing hierarchies, like how a specific diagnosis code nests under a broader disease category. HCOE uses ICD codes organized by CCS categories and ATC medication classes as its training signal, applying contrastive learning that considers both parent and child relationships in the hierarchy. The claims come from an unreviewed preprint, arXiv:2609.30763, posted 2026-09-28, and have not yet been peer reviewed or independently replicated.\n\nWhy this matters: most clinical language models treat medical codes as just another vocabulary word, ignoring the fact that codes are organized in strict clinical taxonomies. If a hierarchy-aware embedding genuinely improves tasks like mortality and readmission prediction on real hospital data, that is a meaningful step for the unglamorous but important work of making electronic health record systems more reliable. The paper reports HCOE outperforming on MIMIC-IV benchmarks for mortality prediction, readmission prediction, medication recommendation, and rare drug prediction, plus wins on ICD\u002FATC relation prediction and CCS-to-PheCode transfer tasks.\n\nBenchmark wins from a single unreviewed preprint are a promising data point, not a verdict -- healthcare AI has a long history of papers that look great on MIMIC-IV and struggle the moment they meet a messier, real-world hospital dataset.","[\"healthcare-ai\",\"biobert\",\"clinical-nlp\",\"arxiv-preprint\"]","2026-09-28T04:00:00.000Z","2026-09-28T22:27:01.740Z","2026-09-28T22:27:08.756Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"Add clear attribution that this is an unreviewed arXiv preprint (include the arXiv ID\u002Fdate, e.g. arXiv:2609.30763, posted 2026-09), since the draft asserts benchmark-winning claims with no link or date making them independently verifiable.","resolved","ai",[32,33,34,35],"healthcare-ai","biobert","clinical-nlp","arxiv-preprint",[37],{"name":38,"url":39},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.30763",0,{"sections":42},[43,47,52,57,62,67,72,77,82,87,92,97,102,107],{"name":44,"slug":30,"count":45,"latest_published_at":46},"AI",4900,"2026-09-28T17:44:43.000Z",{"name":48,"slug":49,"count":50,"latest_published_at":51},"Security","security",766,"2026-09-28T15:35:23.000Z",{"name":53,"slug":54,"count":55,"latest_published_at":56},"Policy","policy",405,"2026-09-28T17:00:51.000Z",{"name":58,"slug":59,"count":60,"latest_published_at":61},"Deals","deals",269,"2026-09-28T17:42:46.000Z",{"name":63,"slug":64,"count":65,"latest_published_at":66},"Hardware","hardware",191,"2026-09-28T15:45:00.000Z",{"name":68,"slug":69,"count":70,"latest_published_at":71},"Science","science",154,"2026-09-28T13:19:18.000Z",{"name":73,"slug":74,"count":75,"latest_published_at":76},"Consumer Tech","consumer-tech",139,"2026-09-28T17:09:47.000Z",{"name":78,"slug":79,"count":80,"latest_published_at":81},"Software","software",91,"2026-09-25T20:55:00.000Z",{"name":83,"slug":84,"count":85,"latest_published_at":86},"Dev Tools","dev-tools",87,"2026-09-28T16:11:42.000Z",{"name":88,"slug":89,"count":90,"latest_published_at":91},"Startups","startups",80,"2026-09-28T17:50:28.000Z",{"name":93,"slug":94,"count":95,"latest_published_at":96},"General","general",49,"2026-09-28T16:44:57.000Z",{"name":98,"slug":99,"count":100,"latest_published_at":101},"Gaming","gaming",48,"2026-09-25T18:35:21.000Z",{"name":103,"slug":104,"count":105,"latest_published_at":106},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":108,"slug":109,"count":110,"latest_published_at":111},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]