{"paper":{"title":"Evaluating Pretrained Language Semantics for Structured EHR Prediction Tasks","arxiv_id":"2611.17053","primary_cat":"cs.CL","submitted_at":"2026-11-12T16:34:49+00:00","version_compiled":"v1"},"notice":"Every quoted sentence is the authors' text from the arXiv listing abstract. Labels and probabilities are the outputs of classifier-1.13.0 on the question bank abstract-p1-20260923, exactly as returned. No sentence in this record was written by a language model.","schema":"pith.compiled-record/v0","status":"compiled","abstract":{"source":"arXiv listing abstract v1","sentences":[{"i":0,"p":0.94,"id":"s0","own":false,"role":"background","text":"Structured Electronic Health Records (EHR) models typically treat each clinical concept as a dataset-specific token, discarding the natural-language semantics associated with the event.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.06,"main_claim_p":0.03,"quantitative":false,"quantitative_p":0.05,"role_probabilities":{"gap":0.06,"method":0.0,"result":0.0,"background":0.94,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":1,"p":0.58,"id":"s1","own":false,"role":"background","text":"As a result, relationships between concepts must largely be learned from EHR co-occurrence, without directly exploiting any prior semantic knowledge.","hedge":{"p":0.98,"label":"asserted"},"own_p":0.06,"main_claim_p":0.03,"quantitative":false,"quantitative_p":0.09,"role_probabilities":{"gap":0.42,"method":0.0,"result":0.0,"background":0.58,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":2,"p":0.99,"id":"s2","own":true,"role":"contribution","text":"We investigate how pretrained causal language models can be adapted for longitudinal EHR prediction by developing tEHR.","hedge":{"p":0.85,"label":"asserted"},"own_p":0.91,"main_claim_p":0.06,"quantitative":false,"quantitative_p":0.04,"role_probabilities":{"gap":0.0,"method":0.01,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.99,"interpretation":0.0}},{"i":3,"p":0.99,"id":"s3","own":true,"role":"method","text":"Here, we represent events with their natural language description and use this text stream to apply EHR-adaptive continued pretraining, followed by task-specific discriminative fine-tuning using LoRA, utilising the LLM as a longitudinal patient-encoder.","hedge":{"p":0.98,"label":"asserted"},"own_p":0.95,"main_claim_p":0.06,"quantitative":false,"quantitative_p":0.04,"role_probabilities":{"gap":0.0,"method":0.99,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.01,"interpretation":0.0}},{"i":4,"p":0.93,"id":"s4","own":true,"role":"result","text":"We evaluate this approach across four prediction tasks on the MIMIC-IV dataset and pancreatic cancer prediction on the CPRD dataset at 1 to 12 month horizons, with tEHR using the Mistral-7B-v0.3 backbone achieving the strongest overall performance.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.97,"main_claim_p":0.61,"quantitative":true,"quantitative_p":0.98,"role_probabilities":{"gap":0.0,"method":0.05,"result":0.93,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.02,"interpretation":0.0}},{"i":5,"p":1.0,"id":"s5","own":true,"role":"result","text":"Ablations show that performance declines when clinical descriptions are removed or misaligned, when numerical values are omitted, and when either stage of adaptation is removed, while explicit elapsed-time descriptions provide little additional benefit.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.97,"main_claim_p":0.16,"quantitative":true,"quantitative_p":0.69,"role_probabilities":{"gap":0.0,"method":0.0,"result":1.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":6,"p":0.59,"id":"s6","own":true,"role":"interpretation","text":"Analysis of the learned patient representations further reveals clinically coherent structure associated with distinct patterns of disease risk and clinical history.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.95,"main_claim_p":0.14,"quantitative":false,"quantitative_p":0.04,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.41,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.59}}]},"compiled":{"main_claim":{"p":0.99,"text":"We evaluate this approach across four prediction tasks on the MIMIC-IV dataset and pancreatic cancer prediction on the CPRD dataset at 1 to 12 month horizons, with tEHR using the Mistral-7B-v0.3 backbone achieving the strongest overall performance.","location":"abstract s4","hedge_strength":{"p":1.0,"label":"asserted"}},"provenance":{"route":{"via":"TypeSafe API","model":"classifier-1.13.0","endpoint":"https://api.typesafe.ai/v1/systemone","provider":"TypeSafe"},"packing":"one request per paper","requests":1,"input_tokens":5736,"cost_usd_reported":0.0},"compiled_at":"2026-11-14","bank_version":"abstract-p1-20260923","model_version":"classifier-1.13.0","abstract_states":{"claims_first":0.05,"states_limitation":0.87,"human_participants":0.78,"controlled_comparison":0.44,"randomized_assignment":0.01,"states_quantitative_result":0.38,"compares_with_prior_results":0.46,"mentions_code_or_data_availability":0.03},"evidential_basis":{"p":0.66,"value":"observation or experiment","probabilities":{"argument":0.0,"literature":0.0,"proof_or_derivation":0.0,"analysis_of_existing_data":0.24,"observation_or_experiment":0.66,"simulation_or_computation":0.1}},"contribution_type":{"p":0.99,"value":"method or tool","probabilities":{"review":0.0,"method_or_tool":0.99,"empirical_study":0.01,"simulation_study":0.0,"theoretical_result":0.0,"dataset_or_benchmark":0.0,"position_or_commentary":0.0}},"contribution_statement":{"p":0.75,"text":"Here, we represent events with their natural language description and use this text stream to apply EHR-adaptive continued pretraining, followed by task-specific discriminative fine-tuning using LoRA, utilising the LLM as a longitudinal patient-encoder.","location":"abstract s3"}},"coverage":{"tier":"abstract","levels_run":["P abstract"],"not_examined":["full text","figures and tables","references and citing sentences","relations to other papers"],"sources_read":["arXiv listing abstract v1"],"levels_not_run":["S sections","T body sentences","C citations","R cross-paper relations"],"sentences_examined":7,"sentences_in_abstract":7},"measured_accuracy":{"url":"https://pith.science/accuracy","measured_on":"2026-10-02","method":"two independent model families labelled a fresh random sample blind; the figure is how often the published label matches their consensus","sentence_role_agreement":0.92,"main_claim_agreement":0.908,"typed_citation_agreement":0.933,"typed_citation_agreement_95_interval":[0.917,0.945],"review_comments_accurate":0.758,"review_comments_accurate_95_interval":[0.684,0.82],"review_caught_admitted_error_found_or_related":0.448,"review_comments_judged":{"judged_by":["gpt-5.6-luna","grok-4.7"],"judged_against":"Pith's own text copy of each paper: text extracted from the PDF, the first 90,000 characters","measured_on":"2026-10-04","accurate":113,"decided":149,"sampled_reviews":200,"split_by_lane":false,"by_lane":"https://pith.science/accuracy#by-lane"}}}