What question did this study set out to answer?

This research aims to develop a deep learning model that predicts pediatric sleep apnea using ECG and SpO2 data.

January 10, 2026Measurement0 citationsOpen Access

Combined explainable deep learning model to predict pediatric sleep apnea from ECG and SpO2

CGClara García-VicenteUniversidad de Valladolid GGGonzalo C. Gutiérrez‐TobalUniversidad de Valladolid FVFernando Vaquerizo-VillarUniversidad de Valladolid

Key Points

This research aims to develop a deep learning model that predicts pediatric sleep apnea using ECG and SpO2 data.
Developed an explainable deep learning model

Structured PICO

Does an explainable deep learning model using overnight ECG and SpO2 signals accurately predict pediatric sleep apnea severity?

Population

3,320 pediatric patients aged 0 to 13 years undergoing sleep studies for suspected obstructive sleep apnea, pooled from three databases (CHAT, PATS, and UofC).

Intervention

Explainable deep learning (DL) model based on a stacked ensemble strategy integrating convolutional neural networks (CNNs) fed with overnight single-lead ECG and SpO2 recordings.

Comparator

Standard overnight polysomnography (PSG) with manual scoring of the apnea-hypopnea index (AHI) by sleep specialists.

Outcome

Apnea-hypopnea index (AHI) estimation and 4-class pediatric OSA severity classification (no OSA, mild, moderate, severe).surrogate

An explainable deep learning model integrating overnight ECG and SpO2 signals can estimate pediatric sleep apnea severity, potentially simplifying diagnosis compared to full polysomnography.

Abstract

• First interpretable DL model using SpO 2 and ECG to assess OSA severity in children. • Improved diagnostic ability of pediatric OSA over previous DL approaches. • XAI offered visual and quantitative insights on cardiorespiratory OSA patterns • SpO 2 is more relevant in moderate and severe cases, and ECG in mild and no OSA. • Findings highlight the complementary nature of SpO 2 and ECG for enhanced diagnosis. Combining deep learning (DL) with eXplainable Artificial Intelligence (XAI) techniques has led to clinically applicable models that simplify the diagnosis of pediatric obstructive sleep apnea (OSA) using a restricted number of cardiorespiratory signals. However, no prior study has applied these techniques to concurrently analyze electrocardiogram (ECG) and oxygen saturation (SpO 2 ) data. Here, we present an explainable DL approach integrating convolutional neural networks with overnight SpO 2 and ECG signals to identify pediatric OSA. SHapley Additive exPlanations (SHAP) XAI technique was used to extract relevant patterns linked to pediatric OSA and explain the model decisions. Patients ( n = 3,320) from the semi-public Childhood Adenotonsillectomy Trial (CHAT) and Pediatric Adenotonsillectomy Trial for Snoring (PATS), and the private University of Chicago (UofC) databases were analyzed. Performance obtained Cohen’s 4-class kappa of 0.549, 0.457, and 0.378 in CHAT, PATS, and UofC, respectively. Shapley values increased with OSA severity and highlighted the complementarity of SpO 2 and ECG, with SpO 2 being more relevant in moderate and severe cases and ECG in mild or no OSA cases. SHAP visualizations identified SpO 2 desaturations linked to clusters of apneic events and those occurring independently. It also highlighted bradycardia-tachycardia and ECG cardiovascular risk patterns, including variations in P and T waves, PQ and QT intervals, and the QRS complex. Shapley values identified correlations between respiratory and cardiac patterns, showing that desaturations in OSA are linked to cardiac changes. Therefore, our interpretable DL approach may improve pediatric OSA diagnosis by integrating breathing information and accompanying cardiac changes, supporting its effective adoption in clinical settings.

Ask AI

Helpful

Bookmark

View Full Paper