Generative Artificial Intelligence Use and Applied Higher-Order Cognitive Performance Among Health Informatics Students: A Multicenter Cross-Sectional Case-Based Assessment Study
Related Experiment Videos
Anas Ali Alhur1, Nouf Khaled Al-Kahtani1, Emad Abdul-Mahsoud Mahgoub2
1Department of Health Information Management and Technology, College of Public Health, Imam Abdulrahman bin Faisal University, Dammam, Saudi Arabia.
Background:
Generative artificial intelligence (GenAI) is widely used by university students, but evidence linking routine use with demonstrated higher-order cognitive performance is limited, particularly in Health Informatics education.
Purpose:
To examine associations between GenAI use and applied higher-order cognitive performance among undergraduate Health Informatics students in Saudi Arabia.
Methods:
In this multicenter cross-sectional study, a convenience sample of 431 students from seven Saudi universities completed an online questionnaire (May-July 2026). GenAI use was self-reported (duration, frequency, weekly hours, academic-use intensity, responsible use, and formal AI training). Performance was measured with a newly developed, expert-reviewed, 30-item case-based assessment of critical thinking, problem-solving, and decision-making (weighted 0-3 scoring; total 0-90), completed remotely without proctoring. Nonparametric tests and hierarchical regression with heteroscedasticity-robust (HC3) standard errors and university fixed effects were used.
Results:
Of 431 participants, 425 (98.6%) reported prior GenAI use. The mean total score was 57.14/90 (SD 15.41). Frequency, duration, and weekly hours of GenAI use were not associated with performance. Academic GenAI-use intensity showed a weak unadjusted inverse correlation (ρ = -0.15) but no association after adjustment for university (B = 0.10; 95% CI -1.61 to 1.80). Adding university fixed effects increased R 2 from 0.080 to 0.287. Formal AI training was negatively associated with performance (B = -3.48; 95% CI -6.40 to -0.56).
Conclusion:
In this convenience sample, routine GenAI exposure was not independently associated with performance on a newly developed case-based assessment, whereas scores varied considerably between universities for reasons this study could not identify. Because the design was cross-sectional, exposure was self-reported, and independent completion could not be verified, the findings are associational and the instrument requires external validation. Health Informatics programs may benefit from attending to how students use GenAI rather than how often.