Gender Differences in Depression Symptom Severity: An AI-expedited Meta-analysis

This meta-analysis aims to examine gender differences in depression symptoms across various age groups and specific population samples, using an automated review approach powered by large language models (LLMs). Articles were retrieved using a PubMed-based pattern-matching tool, followed by OpenAI LLM inferences to identify studies reporting gender differences in depression symptoms. Studies were included if LLM reported that they contain mean values for depression symptoms, standard deviations, and sample sizes for both genders, allowing for meta-analysis of standardized mean differences (SMDs). Across all age groups, our results (SMD = 0.24 [0.23, 0.25]) were comparable to an earlier meta-analysis by Salk et al. (SMD = 0.27 [0.26, 0.29]). In addition, we found persistent gender differences in specific populations, such as cancer patients, and disparity variability depending on the measurement instrument used. Our findings confirm gender differences in depression symptoms using new data and support the utility of LLM-driven meta-analyses.