Healthcare-associated infections (HAI) remain a major challenge for healthcare systems, despite decades of prevention efforts and increasingly detailed surveillance definitions. Conditions such as central line–associated bloodstream infection, catheter-associated urinary tract infection, and surgical site infection are largely preventable, yet their detection and reporting rely heavily on manual review by infection prevention teams. This process is time-consuming, subject to interpretation, and prone to variability between institutions, which can undermine confidence in reported metrics and limit the effectiveness of prevention programs.
Advances in artificial intelligence (AI), particularly generative AI and large language models, offer new opportunities to improve HAI surveillance. Unlike traditional automated approaches that depend mainly on structured data, generative AI can interpret unstructured clinical text and integrate complex clinical information in a way that resembles expert human review. Early studies using simulated cases and retrospective patient records suggest that these tools can identify HAI criteria with accuracy comparable to experienced infection preventionists, while substantially reducing the time required for case review. By standardizing interpretation and minimizing subjective judgment, AI-based surveillance could reduce inter-hospital variability and improve the consistency of reporting.
From a clinical and operational perspective, the potential benefits are considerable. Automating surveillance may allow infection prevention teams to redirect their efforts toward active prevention, education, and frontline engagement, rather than spending a large proportion of time on chart review. At the same time, more objective and timely detection could strengthen trust in HAI metrics and support the development of alternative outcome measures based entirely on electronic health record data. However, important challenges remain, including data quality, generalizability across healthcare settings, algorithmic bias, and the need for prospective validation. Careful integration into clinical workflows and strong governance will be essential to ensure that AI enhances, rather than replaces, expert oversight. With responsible implementation, AI has the potential to transform HAI surveillance into a more efficient, transparent, and prevention-oriented process.
Reference
Morgan DJ, Goodman KE, Branch-Elliman W, et al. Using generative AI for healthcare-associated infection surveillance. Clinical Infectious Diseases, 2025
