Class Imbalance and Batch Effects in LLM-Based Screening for Systematic Reviews
arXiv:2608.14737
Abstract
This study analyses LLMs in imbalanced binary classification, using study screening in systematic reviews as the application domain. An experiment was conducted in five reviews, comparing individual and batch processing, with and without prevalence metadata. The results indicate a limited influence of the prevalence metadata, with no evidence that it improves performance. In contrast, batch processing produced larger behavioral changes that varied according to the prevalence of the class. The aggregate and item-level analyses did not always coincide. Therefore, batch processing should be evaluated not only in terms of cost, but also in relation to its effects on decision-making behavior.
12 pages, 4 figures. Accepted at ENIAC 2026 (National Meeting on Artificial and Computational Intelligence), part of BRACIS 2026