TITLE:
Using Deep Learning to Predict QT Prolongation in ICU Patients on Antipsychotic Therapy: A Case Study
AUTHORS:
Rocco de Filippis, Abdullah Al Foysal
KEYWORDS:
QT Prolongation, Antipsychotic Therapy, Deep Learning, LSTM Networks, Cardiac Risk Prediction
JOURNAL NAME:
Open Access Library Journal,
Vol.12 No.1,
January
21,
2025
ABSTRACT: QT prolongation is a significant cardiac risk factor, particularly in ICU patients undergoing antipsychotic therapy, where it can lead to life-threatening arrhythmias such as torsades de pointes or sudden cardiac arrest. This study presents a robust deep-learning approach using Long Short-Term Memory (LSTM) networks to predict QT prolongation based on clinically relevant features, including demographics, electrolyte levels, heart rate, and medication types. A synthetic dataset mimicking ICU patient profiles was used for model training and evaluation, achieving high accuracy, precision, and recall. The model demonstrated strong discriminatory power, with an Area under the Curve (AUC) of 0.94, and effectively balanced sensitivity and specificity, ensuring reliable predictions for both prolonged and normal QT intervals. Performance metrics, including ROC and Precision-Recall curves, and a confusion matrix validated the model’s robustness and generalizability. Key strengths include minimal overfitting and adaptability for real-time ICU deployment. While the use of synthetic data provided a controlled environment for evaluation, validation on real-world datasets is necessary for clinical adoption. This study highlights the potential of integrating deep learning models into ICU workflows to enable early detection of QT prolongation, guide personalized treatment strategies, and improve patient outcomes.