{"id":"W4414702782","doi":"10.1038/s41746-025-02004-3","title":"Large language models forecast patient health trajectories enabling digital twins","year":2025,"lang":"en","type":"article","venue":"npj Digital Medicine","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":24,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Canadian Institutes of Health Research; National Institutes of Health; Genentech; IXICO; H. Lundbeck A/S; Servier; Eisai; Northern California Institute for Research and Education; Pfizer; Novartis Pharmaceuticals Corporation; Biogen; F. Hoffmann-La Roche; Roche; Eli Lilly and Company; BioClinica; European Commission; Alzheimer's Disease Neuroimaging Initiative; Meso Scale Diagnostics; National Institute on Aging; Alzheimer's Association; Foundation for the National Institutes of Health","keywords":"Benchmarking; Health care; Health records; Normalization (sociology); Imputation (statistics); Generative grammar; Database normalization; Missing data","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001906256,0.0002153288,0.0004554494,0.000273842,0.0001704624,0.00007531745,0.00009685022,0.0001002751,0.0000811533],"category_scores_gemma":[0.0007389533,0.0001677839,0.00008589635,0.0005566954,0.0001295529,0.0005450612,0.00004652438,0.0002698581,0.0000401977],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002599398,"about_ca_system_score_gemma":0.0004940492,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003137581,"about_ca_topic_score_gemma":0.0000703808,"domain_scores_codex":[0.9980046,0.00001657166,0.0007009477,0.0003561802,0.0003859248,0.0005358416],"domain_scores_gemma":[0.9988022,0.0001953057,0.0001251379,0.0003512859,0.0002151814,0.0003108541],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"qualitative","study_design_scores_codex":[0.0002238129,0.0006433396,0.01202997,0.0005875958,0.00009170426,0.00003842965,0.04241237,0.0000275152,0.0001001391,0.007729637,0.01590261,0.9202129],"study_design_scores_gemma":[0.00282364,0.01202205,0.005954001,0.01122834,0.0002959819,0.0003250997,0.533088,0.01004221,0.007455267,0.05759849,0.3575356,0.001631337],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9030238,0.004197408,0.009038663,0.02100214,0.001688997,0.001018943,0.0001178372,0.0002662983,0.05964588],"genre_scores_gemma":[0.9929034,0.00009853057,0.00004811145,0.003797984,0.0005977625,0.00003375176,0.0003628409,0.00002467875,0.002133007],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9185815,"threshold_uncertainty_score":0.6842029,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.09066821201173551,"score_gpt":0.3957070671382421,"score_spread":0.3050388551265066,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}