Background: Nurses working in high-pressure environments, such as intensive care units (ICUs), frequently face interruptions while using nursing information systems (NIS). These interruptions significantly impact their cognitive load and task accuracy. Our previous research, utilizing event-related potential (ERP) technology, revealed that interruptions disrupt nurses’ working memory and reduce task resumption efficiency. Moreover, slower resumption times after interruptions are associated with worse task performance. Building on these findings, we propose the development of an “Intelligent Task Recovery Assistant” module to support nurses in quickly and accurately resuming interrupted tasks. Objective: To design an intelligent “Task Recovery Assistant” module that records, tracks, and provides smart prompts to facilitate efficient task resumption following interruptions. Additionally, the study aims to preliminarily explore the effects of this module on nurses’ performance in NIS tasks under different types of interruptions. Methods: A parallel-controlled experimental study was conducted using eye-tracking technology. A convenience sample of 20 nursing students from medical colleges in Beijing was recruited, with 10 students assigned to the control group and 10 to the intervention group. Participants were required to complete five data-entry tasks within the NIS, including vital signs entry, intake and output recording, pressure injury risk assessment, fall risk assessment, and unplanned extubation assessment. During each task, participants were subjected to either task-related or task-unrelated interruptions. Task-unrelated interruptions were triggered by researchers asking basic nursing knowledge questions, while task-related interruptions required participants to check patient information by navigating to a different system interface before resuming the main task. The intervention group used the NIS with the “Task Recovery Assistant” module, while the control group used the standard NIS. The “Task Recovery Assistant” module featured two core functions. Task Recovery Prompt: If no operation was detected for 8 s, the system would automatically highlight the last interaction point. Task Jump Function: This function allowed users to switch to other patient information interfaces and return to the main task interface with a single click. Upon returning, the system automatically highlighted the last interaction point to remind users where they left off. Eye-tracking indicators, including fixation count, fixation duration, pupil diameter, and saccades in the areas of interest (AOIs), were observed before and after interruptions. Results: For the vital signs entry task, there were no statistically significant differences in eye-tracking indicators between the intervention and control groups before interruptions. After interruptions, the intervention group showed a significantly shorter cumulative fixation duration (6922.0 ± 3824.4 ms vs 12347.1 ± 9609.0 ms, P = 0.005, t = -1.659) and fewer saccades (10.1 ± 6.6 vs 23.6 ± 21.7, P = 0.001, t = -1.881) in the task area compared to the control group. Additionally, the first complete fixation duration in the intervention group was significantly longer than that in the control group (322.6 ± 351.2 ms vs 142.5 ± 72.8 ms, P = 0.023, t = -1.588). For the other four tasks, no significant differences in eye-tracking indicators were observed between the intervention and control groups. Conclusion: The “Task Recovery Assistant” addresses a critical need for interruption management in nursing information systems by incorporating cognitive science principles and intelligent design. It significantly improves task resumption efficiency for multi-data entry tasks such as vital signs entry. Future work will focus on optimizing the module to support other NIS tasks, thereby enhancing overall task performance.