Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Hybrid Enumeration Framework for Optimal Counterfactual Generation in Post-Acute COVID-19 Heart Failure

Jingya Cheng, Alaleh Azhir, Jiazi Tian, Hossein Estiri

TL;DR

This work addresses the need for individualized risk assessment and intervention planning for post-acute sequelae of COVID-19 in patients with pre-existing heart failure. It introduces a hybrid counterfactual framework that combines regularized predictive modeling with both exhaustive enumeration for sparse binary features and optimization-based search (NICE and MOC) to explore feasible, patient-specific interventions, leveraging TLDR-derived temporal features from a large EHR cohort. The approach achieves strong discrimination (AUROC 0.88, 95% CI 0.84–0.91) and produces interpretable, patient-level counterfactuals illustrating how modifying comorbidity patterns or treatments could reduce PASC-related HF admissions. This unified pipeline offers deterministic optimality where tractable and scalable approximations otherwise, with potential applications in in-silico trials and drug repurposing to inform personalized preventive strategies in complex biomedical settings.

Abstract

Counterfactual inference provides a mathematical framework for reasoning about hypothetical outcomes under alternative interventions, bridging causal reasoning and predictive modeling. We present a counterfactual inference framework for individualized risk estimation and intervention analysis, illustrated through a clinical application to post-acute sequelae of COVID-19 (PASC) among patients with pre-existing heart failure (HF). Using longitudinal diagnosis, laboratory, and medication data from a large health-system cohort, we integrate regularized predictive modeling with counterfactual search to identify actionable pathways to PASC-related HF hospital admissions. The framework combines exact enumeration with optimization-based methods, including the Nearest Instance Counterfactual Explanations (NICE) and Multi-Objective Counterfactuals (MOC) algorithms, to efficiently explore high-dimensional intervention spaces. Applied to more than 2700 individuals with confirmed SARS-CoV-2 infection and prior HF, the model achieved strong discriminative performance (AUROC: 0.88, 95% CI: 0.84-0.91) and generated interpretable, patient-specific counterfactuals that quantify how modifying comorbidity patterns or treatment factors could alter predicted outcomes. This work demonstrates how counterfactual reasoning can be formalized as an optimization problem over predictive functions, offering a rigorous, interpretable, and computationally efficient approach to personalized inference in complex biomedical systems.

A Hybrid Enumeration Framework for Optimal Counterfactual Generation in Post-Acute COVID-19 Heart Failure

TL;DR

This work addresses the need for individualized risk assessment and intervention planning for post-acute sequelae of COVID-19 in patients with pre-existing heart failure. It introduces a hybrid counterfactual framework that combines regularized predictive modeling with both exhaustive enumeration for sparse binary features and optimization-based search (NICE and MOC) to explore feasible, patient-specific interventions, leveraging TLDR-derived temporal features from a large EHR cohort. The approach achieves strong discrimination (AUROC 0.88, 95% CI 0.84–0.91) and produces interpretable, patient-level counterfactuals illustrating how modifying comorbidity patterns or treatments could reduce PASC-related HF admissions. This unified pipeline offers deterministic optimality where tractable and scalable approximations otherwise, with potential applications in in-silico trials and drug repurposing to inform personalized preventive strategies in complex biomedical settings.

Abstract

Counterfactual inference provides a mathematical framework for reasoning about hypothetical outcomes under alternative interventions, bridging causal reasoning and predictive modeling. We present a counterfactual inference framework for individualized risk estimation and intervention analysis, illustrated through a clinical application to post-acute sequelae of COVID-19 (PASC) among patients with pre-existing heart failure (HF). Using longitudinal diagnosis, laboratory, and medication data from a large health-system cohort, we integrate regularized predictive modeling with counterfactual search to identify actionable pathways to PASC-related HF hospital admissions. The framework combines exact enumeration with optimization-based methods, including the Nearest Instance Counterfactual Explanations (NICE) and Multi-Objective Counterfactuals (MOC) algorithms, to efficiently explore high-dimensional intervention spaces. Applied to more than 2700 individuals with confirmed SARS-CoV-2 infection and prior HF, the model achieved strong discriminative performance (AUROC: 0.88, 95% CI: 0.84-0.91) and generated interpretable, patient-specific counterfactuals that quantify how modifying comorbidity patterns or treatment factors could alter predicted outcomes. This work demonstrates how counterfactual reasoning can be formalized as an optimization problem over predictive functions, offering a rigorous, interpretable, and computationally efficient approach to personalized inference in complex biomedical systems.
Paper Structure (14 sections, 5 equations, 1 figure, 1 table, 1 algorithm)

This paper contains 14 sections, 5 equations, 1 figure, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Method
  3. Hybrid Counterfactual Generation Framework
  4. Binary Feature Identification and Enumeration
  5. Optimization-Based Fallback
  6. NICE (Nearest Instance Counterfactual Explanations).
  7. MOC (Multi-Objective Counterfactuals).
  8. Algorithm Summary
  9. Results
  10. Discussion
  11. Data Availability
  12. Ethics approval
  13. Declaration of Interests
  14. Acknowledgments

Figures (1)

  • Figure 1: (a) ROC for the test set with Youden’s threshold indicated; AUROC = 0.88 (95% CI: 0.84–0.91). (b) Top model predictors ranked by relative importance. (c) Patient-level counterfactuals illustrating PASC-HF risk modification following COVID-19 infection. Case 1 (blue): 83-year-old female with poorly controlled hypertension (HTN) and heart failure with preserved ejection fraction (HFpEF); counterfactual scenario “no HTN” reduces predicted risk from 74% to 8%. Case 2 (pink): 79-year-old female with HTN, HFpEF, coronary artery disease (CAD), and chronic kidney disease (CKD; Cr 1.3–1.8); counterfactual “no CKD” decreases risk from 74% to 34%. Case 3 (green): 54-year-old male with HTN and poorly controlled diabetes (A1c 9.3%), complicated by retinopathy, nephropathy, neuropathy, and diabetic ketoacidosis (DKA); counterfactual “no diabetes” lowers risk from 72% to 15%. Counterfactual analyses assume all other clinical features remain constant.