We combined mining of adverse event reports, corroboration in electronic health records, and experimental validation using single-cell patch clamp to discover andvalidate a QT-DDI between ceftriaxone and lansoprazole. We used a data-driven profile of side effects that are predictive of LQTS to prioritize drug pairs in FAERS. Wecorroborated these findings in the electronic health records by comparing the QTc intervals of patients administered the prioritized drug pair to patients exposed toeither drug alone. We then validated our top prediction (ceftriaxone/lansoprazole) by measuring the dose-dependent changes in hERG channel current using patch-clamp electrophysiology.