Abstract P1-05-36: AI-Based Smartphone App Using a Single-Lead ECG for Automated QTc Diagnostics in Oncology

Introduction: Long QT syndrome is a common cardiotoxic side effect of various anti-tumor drugs. Previous cardiological monitoring of oncological patients is primarily complex and requires for non-internal oncologists a consultation. Therefore, the QTc-Tracker smartphone APP was developed, which enabled a tele-cardiological diagnosis of the QTc time with standard single-lead ECG devices. As a result, diagnosis times could already be reduced by 99%. The further development examined an automatic determination of the QT time using the smartphone APP. However, since single-lead ECG devices are significantly more susceptible to interference, the determination of the QT time is more complex than with 12-lead ECGs. Methods: The QTc-Tracker smartphone APP was developed to determine the QT time. Self-tracker single-lead ECG devices were used to record the lead I signal. The ECG recordings were analyzed in the APP and passed on to an external cardiologist as reference. The APP used artificial intelligence and was trained in the first phase and validated in the second phase. The first phase aimed to improve QT time detection. The results of the APP were compared with the findings of the external cardiologist. In both phases, ECGs from breast cancer patients receiving ribociclib were used. Results: A total of 1889 single-lead ECGs were carried out. 248 of these could not be evaluated (13%). QTc prolongation, according to CTCAE, was diagnosed in 41 cases (2.5%). 878 of the evaluable ECGs were used for the training phase and 763 for the evaluation phase. In the first group (before the improvement), the sensitivity to automatically detect a prolongation of the QT time was 36%, and the specificity was 96%. In the evaluation collective (after the training), the sensitivity went up to 85%, and the specificity was unchanged at 96%. Conclusions: The trained method of the QTc tracker is able to reliably detect a QT time lengthening even without a cardiological diagnosis only by using single-lead self-tracker ECG’s. In the rare cases in which an elongation was not detected, the cardiac diagnosis was only a few milliseconds above the threshold value. This artificial intelligence-based smartphone APP is not intended to replace the cardiological diagnosis, but it can simplify routine processes and help to decide which patients need a cardiological examination more urgently.

Citation Format: Timo Schinköthe, Christian Horst Tonk, Nadia Harbeck, Vanda Carmelo, Joana Gomes Feliciano, Rachel Wuerstlein, Sherko Küemmel, Annette Schmidt. AI-Based Smartphone App Using a Single-Lead ECG for Automated QTc Diagnostics in Oncology [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P1-05-36.