Id: 4349991 Multi-modal Artificial Intelligence Algorithm for the Prediction of Left Atrial Low-voltage Areas in Atrial Fibrillation Patient Based on Sinus Rhythm Electrocardiogram and Clinical Characteristics: a Retrospective, Multicenter Study
Year: 2025
Type: article
Abstract: alerts).Developing automated systems to identify transmissions not requiring clinical attention could substantially reduce inefficiencies and improve quality of care.Objective: This study aims to develop and validate a machine based model, with artificial intelligence, that accurately predicts which ILR transmissions may be automatically archived -based on criteria and adjudication by technicians who triage these reports.Methods: We analyzed 92,335 transmissions from ILR devices by Medtronic from 8 different cardiology clinics over 16 months.Transmissions were labeled as 'not-relevant' if triaging technicians archived them, and 'relevant' if they processed them for physician review.A gradient boosting classifier was trained using a time-series cross-validation approach to maintain temporal integrity.We employed an LLM model to extract features from patients' clinical history.Feature engineering incorporated temporal patterns, historical data, and episode-level aggregations.Results: The gradient boosting model achieved an AUROC of 0.95.The model correctly identified 97.3%% of relevant alerts, despite their relative rarity (5.9%).The model correctly identified 42% of not-relevant transmissions.Further review by clinicians revealed that only 0.66% out of the 2.7% miss-classified not-relevant alerts required prompt attention by clinicians.Conclusion: AI models can be trained to automatically detect transmissions that would be archived by triaging technicians.Training these models with triage adjudication by professional clinicians instead of technicians, as well as AI based extraction of other features within alert episodes, can reduce false and increase the accuracy of ILR remote monitoring.
Source: Heart Rhythm O2
Authors: Yirao Tao, Xingpeng Liu, Deyun Zhang
Language: English
Topic: ECG Monitoring and Analysis
Subfield: Cardiology and Cardiovascular Medicine
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $2,150
OpenAlex ID: W4413963552