---
title: "When Reconstruction Fails: Doubly Robust Mediation with Multimodal Representations"
id: "medrxiv-13-when-reconstruction-is-not-enough-doubly-robust-counterfactual-and"
canonical_url: "https://medichelpline.com/clinical-feed/medrxiv-13-when-reconstruction-is-not-enough-doubly-robust-counterfactual-and"
content_type: "clinical_feed_article"
specialty: "Pharmacology"
source_name: "medRxiv (Clinical Preprints)"
source_url: "https://www.medrxiv.org/content/10.64898/2026.10.21.26366137v1?rss=1"
published_at: "2026-10-23T12:00:00.000Z"
evidence_level: "Verified Feed"
license: "CC-BY-NC-4.0 / Informational Use"
---
# When Reconstruction Fails: Doubly Robust Mediation with Multimodal Representations
## Provenance & Clinical Metadata
- **Canonical URL:** https://medichelpline.com/clinical-feed/medrxiv-13-when-reconstruction-is-not-enough-doubly-robust-counterfactual-and
- **Specialty:** [Pharmacology](https://medichelpline.com/clinical-feed/pharmacology.md)
- **Primary Source:** medRxiv (Clinical Preprints)
- **Source URL:** [Original Journal Publication](https://www.medrxiv.org/content/10.64898/2026.10.21.26366137v1?rss=1)
- **Published At:** 2026-10-23T12:00:00.000Z
- **Evidence Rating:** Verified Feed
## Executive GIST (TL;DR)
- The study examines whether learned **multimodal representations** of high-dimensional health records preserve the information needed for causal identification rather than only optimizing reconstruction. - Investigators defined two families of estimands: counterfactual outcomes under alternative treatments (including phenotype-specific and patient-level counterfactual risk profiles) and interventional direct and indirect effects under exposure-induced intermediate confounding. - The authors proved formal conditions under which a representation preserves identification, and they showed that **interventional mediation** requires stronger preservation than counterfactual comparison. - A theoretical result demonstrated that no training objective that depends solely on covariates—reconstruction, contrastive, or physics-informed—can certify that a representation is sufficient for causal identification. - In a linear structural model, the study team observed that a representation could sacrifice an arbitrarily small share of reconstruction accuracy yet produce arbitrarily large bias in causal estimates. - The authors decomposed cross-fitted **doubly robust** estimator error into a first-order representation bias (not removed by double robustness or cross-fitting) and a second-order remainder from nuisance estimation errors; this leads to deteriorating coverage as sample size grows when representation is insufficient. - They proposed a practical workflow: train modality-specific variational autoencoders outside the estimation sample, use cross-fitted doubly robust and one-step estimators, lock phenotypes, run select-then-confirm pathway analysis, create portable representations for replication, and apply diagnostics for representation insufficiency, exposure leakage, and sensitivity. - Simulation results: when a rare confounder was discarded by the representation and was moderately strong, interventional indirect-effect intervals had 2% coverage at n=3,000; reconstruction loss did not reveal the problem, but a simple predictive diagnostic detected it in 100% of replicates. - Data were simulated; code and implementations are publicly available at the repository provided by the authors.
## Clinical Analysis & Structured Key Points
When Reconstruction Is Not Enough: Doubly Robust Counterfactual and Interventional Mediation Analysis with Learned Multimodal Representations | medRxiv Skip to main content When Reconstruction Is Not Enough: Doubly Robust Counterfactual and Interventional Mediation Analysis with Learned Multimodal Representations View ORCID Profile Cong Cao doi: https://doi.org/10.64898/2026.10.21.26366137 Cong Cao Stanford university Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Cong Cao For correspondence: ccao4541{at}gmail.com Abstract Info/History Metrics Data/Code Preview PDF Abstract Learned representations of high-dimensional multimodal health records are increasingly used to adjust for confounding, yet they are usually trained to reconstruct the data rather than to preserve the information that causal identification requires. We study two families of estimands: counterfactual outcomes under alternative treatments, including phenotype-specific effects and patient-level counterfactual risk profiles of the kind digital twins provide, and interventional direct and indirect effects under exposure-induced intermediate confounding. We give conditions under which a representation preserves identification and show that interventional mediation is more demanding than counterfactual comparison. We prove that no training objective that depends only on the covariates, whether reconstruction, contrastive, or physics-informed, can certify that a representation is sufficient, and in a linear structural model we show that a representation can lose an arbitrarily small share of reconstruction accuracy yet induce arbitrarily large bias. We decompose the error of cross-fitted doubly robust estimators into a first-order representation bias, which neither double robustness nor cross-fitting removes, and a second-order remainder of nuisance-estimation errors, so coverage deteriorates as the sample size grows when the representation is insufficient. We propose a workflow combining modality-specific variational autoencoders trained outside the estimation sample, cross-fitted doubly robust and one-step estimators, locked phenotypes, select-then-confirm pathway analysis, portable representations for external replication, and diagnostics for representation insufficiency, exposure leakage, and sensitivity. In simulations, a representation that discarded a rare confounder gave interventional indirect-effect intervals with 2% coverage at n=3,000 when the confounder was moderately strong, reconstruction loss did not identify the problem, and a simple predictive diagnostic flagged it in 100% of replicates. 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: This study used only simulated data. No human participants or identifiable data were involved, and ethical approval was not required. 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 No patient data were analyzed; all results are based on simulated data. Code implementing the estimators, the diagnostics, and the simulation study is publicly available at https://github.com/congca0/reconstruction-is-not-enough 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 4.0 International license . Back to top Previous Next Posted October 23, 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 When Reconstruction Is Not Enough: Doubly Robust Counterfactual and Interventional Mediation Analysis with Learned Multimodal Representations 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 When Reconstruction Is Not Enough: Doubly Robust Counterfactual and Interventional Mediation Analysis with Learned Multimodal Representations Cong Cao medRxiv 2026.10.21.26366137; doi: https://doi.org/10.64898/2026.10.21.26366137 Share This Article: Copy Citation Tools When Reconstruction Is Not Enough: Doubly Robust Counterfactual and Interventional Mediation Analysis with Learned Multimodal Representations Cong Cao medRxiv 2026.10.21.26366137; doi: https://doi.org/10.64898/2026.10.21.26366137 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 (634) Allergy and Immunology (914) Anesthesia (346) Cardiovascular Medicine (4943) Dentistry and Oral Medicine (490) Dermatology (425) Emergency Medicine (672) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1667) Epidemiology (16164) Forensic Medicine (35) Gastroenterology (1236) Genetic and Genomic Medicine (7213) Geriatric Medicine (747) Health Economics (1091) Health Informatics (5222) Health Policy (1456) Health Systems and Quality Improvement (1812) Hematology (597) HIV/AIDS (1370) Infectious Diseases (except HIV/AIDS) (16427) Intensive Care and Critical Care Medicine (1192) Medical Education (690) Medical Ethics (156) Nephrology (739) Neurology (7499) Nursing (375) Nutrition (1114) Obstetrics and Gynecology (1281) Occupational and Environmental Health (1031) Oncology (3708) Ophthalmology (1079) Orthopedics (401) Otolaryngology (462) Pain Medicine (493) Palliative Medicine (142) Pathology (717) Pediatrics (1844) Pharmacology and Therapeutics (749) Primary Care Research (781) Psychiatry and Clinical Psychology (6065) Public and Global Health (9915) Radiology and Imaging (2499) Rehabilitation Medicine and Physical Therapy (1500) Respiratory Medicine (1263) Rheumatology (655) Sexual and Reproductive Health (794) Sports Medicine (594) Surgery (795) Toxicology (109) Transplantation (315) Urology (303)
## Related Clinical Research

- [Callio Therapeutics hires Chris Hassig as chief scientific officer](https://medichelpline.com/clinical-feed/stat-news-6-stat-up-and-down-the-ladder-the-latest-comings-and-goings.md)
- [U.S. Treasury Proposes Restricting Investment in Chinese Biotech Under COINS Rule](https://medichelpline.com/clinical-feed/stat-news-6-stat-treasury-proposes-major-restrictions-to-u-s-investment-in-chinese-biotech.md)
- [Reproducibility of Common Laboratory Tests to Improve Synthetic Data for Machine Learning](https://medichelpline.com/clinical-feed/medrxiv-12-quantifying-reproducibility-of-common-laboratory-tests-in-a-tertiary-care.md)
- [Medetomidine Appears in Southern California **Fentanyl** Supply: Prevalence, Concentrations, and C](https://medichelpline.com/clinical-feed/medrxiv-1-characterizing-the-emergence-of-medetomidine-in-southern-california-fentanyl.md)
- [Psychosocial profile in adults with congenital heart disease by documented anxiety disorder history](https://medichelpline.com/clinical-feed/plos-one-21-psychosocial-profile-of-adults-with-congenital-heart-disease-according-to.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.