What is the clinical evidence from this study?

Study design: Cohort. Population: Suspected acute pulmonary embolism (n=220). Intervention: Explainable machine-learning models (Tree-based ensemble models) vs. Simpler classifiers. Primary outcome: In-hospital all-cause mortality.

What question did this study set out to answer?

Evaluate whether explainable machine-learning models can predict in-hospital mortality in adults with suspected pulmonary embolism.

February 10, 2026Open Access

Emergency Department Prediction of In-Hospital Mortality in Suspected Pulmonary Embolism: An Explainable Machine Learning Approach

Q: What does this research mean for the field?

Explainable machine learning models integrating routine emergency department variables with validated risk scores can predict in-hospital mortality in adults evaluated for suspected acute pulmonary embolism. Novelty: ClaimNovelty.NOVEL_FINDING. Consensus alignment: ConsensusAlignment.NEUTRAL.

Key Result

Tree-based ensemble machine learning models demonstrated higher discrimination for predicting in-hospital all-cause mortality in suspected pulmonary embolism compared to simpler classifiers.

Key Points

Evaluate whether explainable machine-learning models can predict in-hospital mortality in adults with suspected pulmonary embolism.
Retrospective review of 220 adults in the ED for suspected acute PE from January 2021 to March 2025.
Utilized demographics, vital signs, arterial blood gas indices, imaging findings, and risk scores for predictions.
Trained seven machine learning algorithms and assessed performance using AUC and SHAP for interpretability.
Tree-based ensemble models outperformed simpler classifiers in predicting mortality.
Key predictors included sPESI, oxygenation/arterial blood gas indices, and malignancy, as identified by SHAP analysis.
Findings were consistent in the subgroup of PE-confirmed cases.

Study Design

Type

Cohort (n=220)

Multicenter

Structured PICO

Do explainable machine-learning models integrating routine ED variables with validated risk scores improve the prediction of in-hospital mortality in adults evaluated for suspected acute pulmonary embolism?

Population

220 consecutive adults evaluated for suspected acute pulmonary embolism (PE) in the emergency department, comprising both PE-confirmed and PE-excluded cases.

Intervention

Explainable machine-learning (ML) models (including tree-based ensemble models) integrating routine ED variables (demographics, vital signs, arterial blood gas indices, imaging/echocardiographic findings) with validated risk scores (Wells, Revised Geneva, and sPESI).

Comparator

Simpler classifiers and standard risk scores alone.

Outcome

In-hospital all-cause mortality.hard clinical

Explainable machine learning models combining routine emergency department variables with established risk scores can effectively predict in-hospital mortality in patients with suspected pulmonary embolism.

Limitations

Single-center retrospective design
Requires external multicenter validation and prospective impact studies before clinical implementation

Abstract

Background: Pulmonary embolism (PE) is a significant cause of cardiovascular mortality, and emergency department (ED) management requires early risk assessment to guide monitoring and disposition. Because key decisions are often needed while diagnostic evaluation is ongoing, the simplified Pulmonary Embolism Severity Index (sPESI) may provide limited discrimination for in-hospital outcomes. We evaluated whether explainable machine-learning (ML) models integrating routine ED variables with validated risk scores can predict in-hospital mortality in adults evaluated for suspected acute PE. Methods: A retrospective single-center cohort study was performed, including 220 consecutive adults evaluated for suspected acute PE in the ED between January 2021 and March 2025, comprising both PE-confirmed and PE-excluded cases. Predictors included demographics, vital signs, arterial blood gas indices, available imaging/echocardiographic findings, and Wells, Revised Geneva, and sPESI scores. Seven ML algorithms were trained and internally evaluated using the area under the receiver operating characteristic curve (AUC) and complementary metrics. Model interpretability was assessed using SHAP (SHAPley Additive exPlanations), and a sensitivity analysis was conducted in the PE-confirmed subgroup. Results: Tree-based ensemble models demonstrated higher discrimination for in-hospital all-cause mortality than simpler classifiers. SHAP analyses consistently highlighted sPESI, oxygenation/arterial blood gas indices, and malignancy as key contributors to mortality risk. Findings were similar in the PE-confirmed sensitivity analysis. Conclusions: Explainable ML models combining established risk scores with routinely collected ED variables may complement risk stratification along the suspected-PE pathway. External multicenter validation and prospective impact studies are warranted before clinical implementation.

Bookmark

View Full Paper