Detection of drug-induced QT Syndrome from ECG using machine learning techniques

Induced QT disorder prediction from ECG report using machine learning technique is a new approach. The process is crucial for preventing heartdiseases and removing a broad range of newly invented drugs' after consumption effects. This paper introduces some specific machine learning approaches for classifying drugs that could be harmful enough to cause QT syndrome investigating some ECG reports. The proposed method uses basic features, such as PR, RR, QRS and a range of other properties found in ECG description of the target group. Regression and classification based machine models have been developed to learn the properties of ECG and test the model on an independent dataset. Among the proposed algorithms gradient boosting regression model performed better (RMSE = 0.3) while bragging learning based classifier have shown 89% accuracy indicating the potential of machine learning based approach in identifying the drugs crucial for QT distortion.