Opportunistic Screening of Wolff-Parkinson-White Syndrome using Single-Lead AI-ECG Mobile System: A Real-World Study of over 3.5 million ECG Recordings in China

Wolff-Parkinson-White (WPW) syndrome, a congenital cardiac conduction abnormality with low prevalence, carries a significant risk of sudden cardiac death. Early identification remains challenging due to screening costs and professional resource scarcity. This retrospective real-world study systematically evaluates an integrated Artificial Intelligence-enabled mobile screening system comprising portable single-lead devices, AI primary screening, and cardiologist review. Analyzing 3,566,626 ECG records from 87,836 individuals between 2019 and 2025, the AI model achieved an AUC of 0.6676 and a specificity of 95.92% in complex real-world signal environments. Despite predictive probability bias inherent in ultra-low prevalence contexts, the model demonstrated stable risk stratification, with high-confidence scores concentrated among true positive individuals. The risk of detecting WPW in AI-positive records was 86.2-fold higher than in AI-negative records. By implementing a human-AI collaborative workflow, the volume of ECGs requiring manual review was reduced by approximately 99.5% compared to universal screening. In an ideal collaborative scenario, an average of only 18 ECGs required review to confirm one WPW case, representing a more than 60-fold increase in screening efficiency. Compared to traditional 12-lead ECGs and electrophysiological studies, this system significantly reduced time and medical costs. Our findings suggest that a risk-stratification-based human-AI collaborative system provides a promising paradigm for the early public health detection of low-prevalence, high-risk arrhythmias.

💡 Research Summary

This retrospective real‑world study evaluated a fully integrated, AI‑enabled mobile electrocardiogram (ECG) screening system for opportunistic detection of Wolff‑Parkinson‑White (WPW) syndrome using over 3.5 million single‑lead ECG recordings collected across China from August 2019 to May 2025. The system consists of three components: a portable single‑lead ECG device, a cloud‑based deep‑learning AI model that performs primary risk classification, and a closed‑loop workflow that allows users to request a cardiologist’s review through a smartphone app. After excluding low‑quality recordings, 3,566,626 ECGs from 87,836 individuals remained for analysis.

The AI model flagged 16,457 recordings (0.46 %) as WPW‑positive and the remaining 3,550,169 (99.54 %) as negative. Among the AI‑positive group, 1,984 recordings (12.05 %) were voluntarily sent for cardiologist review; in the AI‑negative group, 46,120 recordings (1.30 %) were similarly reviewed, yielding a total of 48,104 reviewed ECGs (1.35 % of all recordings). The model achieved an area under the receiver‑operating‑characteristic curve (AUC) of 0.6676 and a specificity of 95.92 %, reflecting the challenges of extreme class imbalance (WPW prevalence ≈0.5 %). Sensitivity was modest, but the model demonstrated strong risk‑stratification: true‑positive cases were highly concentrated in the upper 15 % of predicted risk scores, confirming that high‑confidence AI outputs reliably identify the few true WPW cases.

Relative risk analysis showed that an AI‑positive result increased the likelihood of a cardiologist‑confirmed WPW diagnosis by 86.2‑fold compared with AI‑negative results. By integrating AI with human review, the workflow reduced the number of ECGs requiring manual interpretation by approximately 99.5 % relative to universal expert review. The number‑needed‑to‑review (NNR) to confirm one WPW case dropped to 18, representing more than a 60‑fold gain in screening efficiency. In an ideal collaborative scenario, only 18 AI‑positive ECGs would need expert confirmation to identify a single WPW case, dramatically lowering the workload for cardiology services.

Economic evaluation compared the AI‑ECG mobile pathway with conventional in‑hospital 12‑lead ECG and invasive electrophysiology study (EPS). The mobile system shortened the total screening time by 30–36 times and reduced direct medical costs by roughly 37 times, underscoring its potential to alleviate resource constraints in low‑access settings. User‑behavior analysis revealed that individuals who initiated cardiologist review tended to have higher comorbidity burdens (e.g., hypertension, coronary stenting, diabetes), suggesting that AI alerts effectively target higher‑risk populations.

Despite the modest AUC, the study demonstrates that a well‑designed risk‑stratification and human‑AI collaboration can deliver clinically meaningful benefits for a low‑prevalence, high‑risk arrhythmia. The authors propose that extending this framework to other rare channelopathies (e.g., Brugada syndrome, long QT syndrome) and incorporating multi‑lead ECG data could further enhance public‑health screening programs, especially in resource‑limited regions where traditional ECG infrastructure and specialist availability are scarce.