---
title: "ITE-Transformer: Pretrained model for individualized treatment effects from EHRs"
id: "medrxiv-17-the-ite-transformer-a-pre-trained-foundation-model-for-predicting"
canonical_url: "https://medichelpline.com/clinical-feed/medrxiv-17-the-ite-transformer-a-pre-trained-foundation-model-for-predicting"
content_type: "clinical_feed_article"
specialty: "Pharmacology"
source_name: "medRxiv (Clinical Preprints)"
source_url: "https://www.medrxiv.org/content/10.64898/2026.11.19.26368479v1?rss=1"
published_at: "2026-11-20T12:00:00.000Z"
evidence_level: "Verified Feed"
license: "CC-BY-NC-4.0 / Informational Use"
---
# ITE-Transformer: Pretrained model for individualized treatment effects from EHRs
## Provenance & Clinical Metadata
- **Canonical URL:** https://medichelpline.com/clinical-feed/medrxiv-17-the-ite-transformer-a-pre-trained-foundation-model-for-predicting
- **Specialty:** [Pharmacology](https://medichelpline.com/clinical-feed/pharmacology.md)
- **Primary Source:** medRxiv (Clinical Preprints)
- **Source URL:** [Original Journal Publication](https://www.medrxiv.org/content/10.64898/2026.11.19.26368479v1?rss=1)
- **Published At:** 2026-11-20T12:00:00.000Z
- **Evidence Rating:** Verified Feed
## Executive GIST (TL;DR)
- The authors developed the **ITE-Transformer**, a transformer-based foundation model trained on longitudinal electronic health records to predict **individualized treatment effects (ITEs)**. It combines self-supervised representation learning with doubly robust causal machine learning. - Training used structured EHR data from approximately 350,000 participants in the All of Us Research Program; participant-level data are restricted by the program and not publicly shareable. - The framework learns temporally contextualized representations from patients’ diagnosis and medication histories, then predicts treatment assignment, treatment-specific outcome risks, and patient-level ITEs. - Evaluation occurred in two comparative-effectiveness tasks: GLP-1 receptor agonists versus DPP-4 inhibitors in type 2 diabetes with major adverse cardiovascular events as the outcome, and immunotherapy versus chemotherapy in a pan-cancer cohort with all-cause mortality as the outcome. - Investigators reported improved discrimination and calibration for both treatment-assignment and outcome models versus tabular machine-learning baselines and versus a transformer trained without self-supervised pretraining. - The study team observed substantial heterogeneity in predicted ITEs and interpretable associations between predicted treatment response and patients’ diagnoses, medication histories, and comorbidities. - Ethical and data statements: no competing interests declared; LMU Munich noted no local ethical approval was required for this de-identified, non-interventional analysis; All of Us original data collection had IRB approval. - Limitations documented in the source include restricted access to participant-level data under All of Us policies and the preprint status; the source did not report details of certain implementation choices (for example, exact pretraining tasks, hyperparameter settings, or model-size specifications) in the provided abstract.
## Clinical Analysis & Structured Key Points
The ITE-Transformer: A pre-trained foundation model for predicting individualized treatment effects from electronic health records | medRxiv Skip to main content The ITE-Transformer: A pre-trained foundation model for predicting individualized treatment effects from electronic health records Clemente Arceri , Emmanouil Angelis , Philip Heesen , Valentyn Melnychuk , Martin Bendszus , Stefan Bauer , Stefan Feuerriegel doi: https://doi.org/10.64898/2026.11.19.26368479 Clemente Arceri 1 Technical University Munich; Find this author on Google Scholar Find this author on PubMed Search for this author on this site Emmanouil Angelis 1 Technical University Munich; Find this author on Google Scholar Find this author on PubMed Search for this author on this site Philip Heesen 2 University Hospital Heidelberg: UniversitatsKlinikum Heidelberg; Find this author on Google Scholar Find this author on PubMed Search for this author on this site Valentyn Melnychuk 3 LMU Munich Find this author on Google Scholar Find this author on PubMed Search for this author on this site Martin Bendszus 2 University Hospital Heidelberg: UniversitatsKlinikum Heidelberg; Find this author on Google Scholar Find this author on PubMed Search for this author on this site Stefan Bauer 1 Technical University Munich; Find this author on Google Scholar Find this author on PubMed Search for this author on this site Stefan Feuerriegel 3 LMU Munich Find this author on Google Scholar Find this author on PubMed Search for this author on this site For correspondence: feuerriegel{at}lmu.de Abstract Info/History Metrics Data/Code Preview PDF Abstract Transformer-based foundation models have shown strong potential for modeling longitudinal electronic health records (EHRs), yet existing models focus primarily on risk prediction rather than treatment effect estimation, which limits evidence generation. Here, we introduce the ITE-Transformer, a transformer-based framework that combines self-supervised representation learning from longitudinal patient trajectories with doubly robust causal machine learning for predicting individualized treatment effects (ITEs). As a result, our ITE-Transformer allows to study counterfactual questions of how an individual patient's outcome may differ under alternative treatment strategies. The ITE-Transformer learns temporally contextualized representations of patients' diagnosis and medication histories and subsequently uses these representations to predict treatment assignment, treatment-specific outcome risks, and patient-specific ITEs. The ITE-Transformer is trained using structured EHR data from ~350,000 participants in the All of Us Research Program and evaluated in two clinically different comparative-effectiveness tasks: (i) GLP-1 receptor agonists versus DPP-4 inhibitors among patients with type 2 diabetes, with major adverse cardiovascular events as the outcome, and (ii) immunotherapy versus chemotherapy in a pan-cancer cohort, with all-cause mortality as the outcome. Across both applications, the ITE-Transformer improves the discrimination and calibration of treatment assignment and outcome models compared with existing tabular machine learning approaches and an otherwise identical transformer trained without self-supervised pre-training. The ITE-Transformer further identifies substantial heterogeneity in predicted ITEs and yields interpretable patterns of treatment response associated with patients' diagnoses, medication histories, and comorbidities. Together, our work demonstrates how clinical foundation models can be developed for analyzing heterogeneity in treatment effects and thus support real-world evidence generation. Competing Interest Statement The authors have declared no competing interest. Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: No ethical approval was necessary at the host institution (LMU Munich; LMU SOM) due to the de-identified, non-interventional design of the study. The original data collection for the NIH All of Us Research Program received ethical approval from its Institutional Review Board (IRB); see https://www.nih.gov/allofus/about/who-we-are/all-of-us-institutional-review-board I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes Data Availability This study used data from the All of Us Research Program Registered Tier Dataset (version 8), available to authorized users on the All of Us Researcher Workbench. The participant-level data are not publicly available and cannot be shared by the authors under data use policies of the program. https://www.researchallofus.org/data-tools/workbench/ Copyright The copyright holder for this preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license . Back to top Previous Next Posted November 20, 2026. Download PDF Data/Code Email Thank you for your interest in spreading the word about medRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following The ITE-Transformer: A pre-trained foundation model for predicting individualized treatment effects from electronic health records Message Subject (Your Name) has forwarded a page to you from medRxiv Message Body (Your Name) thought you would like to see this page from the medRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share The ITE-Transformer: A pre-trained foundation model for predicting individualized treatment effects from electronic health records Clemente Arceri , Emmanouil Angelis , Philip Heesen , Valentyn Melnychuk , Martin Bendszus , Stefan Bauer , Stefan Feuerriegel medRxiv 2026.11.19.26368479; doi: https://doi.org/10.64898/2026.11.19.26368479 Share This Article: Copy Citation Tools The ITE-Transformer: A pre-trained foundation model for predicting individualized treatment effects from electronic health records Clemente Arceri , Emmanouil Angelis , Philip Heesen , Valentyn Melnychuk , Martin Bendszus , Stefan Bauer , Stefan Feuerriegel medRxiv 2026.11.19.26368479; doi: https://doi.org/10.64898/2026.11.19.26368479 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Areas All Articles Addiction Medicine (642) Allergy and Immunology (928) Anesthesia (355) Cardiovascular Medicine (5059) Dentistry and Oral Medicine (496) Dermatology (432) Emergency Medicine (681) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1703) Epidemiology (16352) Forensic Medicine (37) Gastroenterology (1249) Genetic and Genomic Medicine (7333) Geriatric Medicine (761) Health Economics (1101) Health Informatics (5350) Health Policy (1480) Health Systems and Quality Improvement (1847) Hematology (604) HIV/AIDS (1389) Infectious Diseases (except HIV/AIDS) (16537) Intensive Care and Critical Care Medicine (1210) Medical Education (702) Medical Ethics (157) Nephrology (757) Neurology (7664) Nursing (380) Nutrition (1144) Obstetrics and Gynecology (1308) Occupational and Environmental Health (1046) Oncology (3769) Ophthalmology (1105) Orthopedics (414) Otolaryngology (474) Pain Medicine (508) Palliative Medicine (144) Pathology (727) Pediatrics (1877) Pharmacology and Therapeutics (759) Primary Care Research (793) Psychiatry and Clinical Psychology (6186) Public and Global Health (10051) Radiology and Imaging (2562) Rehabilitation Medicine and Physical Therapy (1540) Respiratory Medicine (1280) Rheumatology (666) Sexual and Reproductive Health (814) Sports Medicine (601) Surgery (809) Toxicology (109) Transplantation (321) Urology (309)
## Related Clinical Research

- [Embolic Stroke and DOAC Failure in Libman-Sacks Endocarditis with Antiphospholipid Syndrome](https://medichelpline.com/clinical-feed/pubmed-43058350.md) (DOI: 10.1007/s00296-026-06331-w)
- [Real-World Safety of Nirmatrelvir Plus Ritonavir for COVID-19 in Japan: Post-Marketing Surveillance](https://medichelpline.com/clinical-feed/pubmed-41621946.md) (DOI: 10.7883/yoken.JJID.2024.380)
- [Amylyx Presents Avexitide Results; MiniMed Launches Tubeless Insulin Pump](https://medichelpline.com/clinical-feed/endocrine-news-0-pharma-friday-november-21-2026.md)
- [Blood methylation profile stratifies 10-year risk for 21 common diseases](https://medichelpline.com/clinical-feed/medrxiv-20-one-blood-methylation-profile-stratifies-ten-year-risk-across-21-common-diseases.md)
- [SNP2Cluster-Py: Context-aware Core SNP clustering for pathogen surveillance](https://medichelpline.com/clinical-feed/biorxiv-3-snp2cluster-py-a-pathogen-agnostic-context-aware-framework-for-core-snp-and.md)

## Navigation
- [← Back to Pharmacology Feed](https://medichelpline.com/clinical-feed/pharmacology.md)
- [← All Clinical Specialties](https://medichelpline.com/clinical-feed.md)
## Medical & Regulatory Disclaimer

> [!CAUTION]
> MedicHelpline content is structured for research, educational, and professional discovery purposes. It does not constitute individual medical advice, clinical diagnosis, or treatment recommendations.
> Always verify dosing, contraindications, and regulatory alerts against official product labeling and primary regulatory sources before clinical decision-making.