Electrocardiograms (ECGs), which capture the electrical activity of the human heart, are widely used in clinical practice for detecting cardiac pathologies, and notoriously difficult to interpret. Many commonly prescribed medications, including antihistamines, antibiotics and antidepressants, can produce a complication known as `drug-induced long QT syndrome (diLQTS)', characterised by a prolongation of the QT-interval on the ECG. It is associated with a life-threatening arrhythmia known as Torsade de Points (TdP)---the leading cause of sudden cardiac death in young, otherwise healthy people. Self-monitoring for diLQTS could therefore save many lives, but detecting it on the ECG is difficult, particularly at high and low heart rates, even for clinicians who routinely read ECGs. Whilst there have been attempts to automate ECG interpretation for several decades, the accuracy of these methods remains limited. In particular, automated QT measurement algorithms have proved unsatisfactory for detecting LQTS. A major challenge for automated QT algorithms is identifying the precise end of the T-wave (the terminal point), especially when the T-wave's morphology is abnormal. This thesis is the first work to examine the ability of laypeople to interpret an ECG for drug-induced QT-prolongation monitoring, devise a novel ECG visualisation technique to enable them to interpret it accurately, and exploit an understanding of the human visual perceptual process to improve automated QT-prolongation detection. The approach draws from the field of pre-attentive processing theory in human vision, showing through several experiments that using pseudo-colour to expose QT-interval duration on the ECG significantly improves laypeople's accuracy in detecting diLQTS at risk of TdP regardless of heart rate and T-wave morphology. An understanding of how humans use pseudo-colour to interpret ECG data is combined with clinical knowledge that considers the morphology of the T-wave to develop a novel, rule-based algorithm that reliably automates the detection of diLQTS, thus facilitating an explainable, shared human-machine ECG interpretation.