Longitudinal clinical notes contain rich evidence of how patients evolve over time, but converting this signal into training supervision for clinical prediction remains challenging. We extend Foresight Learning to clinical prediction by converting time-ordered MIMIC-III notes into examples consisting of past patient context, a natural-language question about a possible future event, and a label resolved from later documentation. This process yields 6,900 prediction examples from 702 admissions across medications, procedures, organ support, microbiology, and mortality. A small LoRA adapter trained on these examples improves over the prompted base model, reducing expected calibration error from 0.1269 to 0.0398 and Brier score from 0.199 to 0.145, while slightly outperforming GPT-5 point estimates on held-out questions. The approach enables reusable clinical prediction supervision from longitudinal notes without hand-engineered structured features or endpoint-specific classifiers.</p>\n","updatedAt":"2026-05-22T03:17:43.368Z","author":{"_id":"631206b4c7722fdac9aa3e34","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/631206b4c7722fdac9aa3e34/QHFDYnlmBs3lPLIKnqQWG.jpeg","fullname":"Ben","name":"Bturtel","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8437448740005493},"editors":["Bturtel"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/631206b4c7722fdac9aa3e34/QHFDYnlmBs3lPLIKnqQWG.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2605.12817","authors":[{"_id":"6a0fca6ba53a61ce2e422d1c","name":"Benjamin Turtel","hidden":false},{"_id":"6a0fca6ba53a61ce2e422d1d","name":"Paul Wilczewski","hidden":false},{"_id":"6a0fca6ba53a61ce2e422d1e","name":"Kris Skotheim","hidden":false}],"publishedAt":"2026-05-12T00:00:00.000Z","submittedOnDailyAt":"2026-05-22T00:00:00.000Z","title":"Training Large Language Models to Predict Clinical Events","submittedOnDailyBy":{"_id":"631206b4c7722fdac9aa3e34","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/631206b4c7722fdac9aa3e34/QHFDYnlmBs3lPLIKnqQWG.jpeg","isPro":false,"fullname":"Ben","user":"Bturtel","type":"user","name":"Bturtel"},"summary":"Longitudinal clinical notes contain rich evidence of how patients evolve over time, but converting this signal into training supervision for clinical prediction remains challenging. We extend Foresight Learning to clinical prediction by converting time-ordered MIMIC-III notes into examples consisting of past patient context, a natural-language question about a possible future event, and a label resolved from later documentation. This process yields 6,900 prediction examples from 702 admissions across medications, procedures, organ support, microbiology, and mortality. A small LoRA adapter trained on these examples improves over the prompted base model, reducing expected calibration error from 0.1269 to 0.0398 and Brier score from 0.199 to 0.145, while slightly outperforming GPT-5 point estimates on held-out questions. The approach enables reusable clinical prediction supervision from longitudinal notes without hand-engineered structured features or endpoint-specific classifiers.","upvotes":2,"discussionId":"6a0fca6ca53a61ce2e422d1f","projectPage":"https://arxiv.org/abs/2605.12817","ai_summary":"Longitudinal clinical notes are converted into temporal prediction examples using Foresight Learning, enabling improved clinical prediction through LoRA adaptation that enhances calibration and reduces uncertainty compared to base models.","ai_keywords":["Foresight Learning","LoRA adapter","clinical prediction","longitudinal notes","temporal prediction","calibration error","Brier score","prompted base model"],"organization":{"_id":"6752078fe8420287f111f953","name":"LightningRodLabs","fullname":"Lightning Rod Labs","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/631206b4c7722fdac9aa3e34/2vSaHnvqO_KKalPzP-md9.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"631206b4c7722fdac9aa3e34","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/631206b4c7722fdac9aa3e34/QHFDYnlmBs3lPLIKnqQWG.jpeg","isPro":false,"fullname":"Ben","user":"Bturtel","type":"user"},{"_id":"69bb66533823efb060256e3b","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/YyBRAtM04DE0dwWGaHLdH.jpeg","isPro":false,"fullname":"지유 전","user":"maruoxi30","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"6752078fe8420287f111f953","name":"LightningRodLabs","fullname":"Lightning Rod Labs","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/631206b4c7722fdac9aa3e34/2vSaHnvqO_KKalPzP-md9.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2605/2605.12817.md"}">
Training Large Language Models to Predict Clinical Events
Published on May 12
· Submitted by Ben on May 22 Abstract
Longitudinal clinical notes are converted into temporal prediction examples using Foresight Learning, enabling improved clinical prediction through LoRA adaptation that enhances calibration and reduces uncertainty compared to base models.
AI-generated summary
Longitudinal clinical notes contain rich evidence of how patients evolve over time, but converting this signal into training supervision for clinical prediction remains challenging. We extend Foresight Learning to clinical prediction by converting time-ordered MIMIC-III notes into examples consisting of past patient context, a natural-language question about a possible future event, and a label resolved from later documentation. This process yields 6,900 prediction examples from 702 admissions across medications, procedures, organ support, microbiology, and mortality. A small LoRA adapter trained on these examples improves over the prompted base model, reducing expected calibration error from 0.1269 to 0.0398 and Brier score from 0.199 to 0.145, while slightly outperforming GPT-5 point estimates on held-out questions. The approach enables reusable clinical prediction supervision from longitudinal notes without hand-engineered structured features or endpoint-specific classifiers.
Community
Longitudinal clinical notes contain rich evidence of how patients evolve over time, but converting this signal into training supervision for clinical prediction remains challenging. We extend Foresight Learning to clinical prediction by converting time-ordered MIMIC-III notes into examples consisting of past patient context, a natural-language question about a possible future event, and a label resolved from later documentation. This process yields 6,900 prediction examples from 702 admissions across medications, procedures, organ support, microbiology, and mortality. A small LoRA adapter trained on these examples improves over the prompted base model, reducing expected calibration error from 0.1269 to 0.0398 and Brier score from 0.199 to 0.145, while slightly outperforming GPT-5 point estimates on held-out questions. The approach enables reusable clinical prediction supervision from longitudinal notes without hand-engineered structured features or endpoint-specific classifiers.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2605.12817 in a model README.md to link it from this page.
Cite arxiv.org/abs/2605.12817 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2605.12817 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.