This research introduces an innovative framework to forecast threshold surpassing in timeseries data by employing survival analysis techniques. Diverging from conventional binary classification methods commonly used for this purpose, our approach provides a distinctive viewpoint, modeling and predicting both the event occurrence and the time-to-event information. This significant departure offers a valuable tool for predicting extreme event occurrences, thus improving decision-makers’ understanding of temporal dynamics concerning such risks and enabling proactive intervention strategies. The efficacy of our approach has been empirically validated using a variety of datasets, including food incident occurrences, web traffic data, and Numenta Artificial datasets. This demonstrates the precision of our method in predicting threshold exceedance events across diverse real-world and simulated scenarios. An comprehensive application in the realm of food safety, leveraging real-world data on food recalls over time, further illustrates the practical usefulness of our approach, particularly in averting and managing high-risk threats such as Salmonella. These findings highlight the broad-ranging implications of our method, especially in contexts where comprehending risk temporal dynamics holds utmost importance.