Computational Neuroscience and Machine Learning Analysis for Understanding Sex-Specific Differences in Microglial Morphology and Cognitive Function in Alzheimer's Disease

Alzheimer’s disease (AD) disproportionately affects women, with females accounting for approximately two-thirds of all dementia cases (Rajan et al., 2021). Despite this striking disparity, the influence of biological sex on AD pathology and behavioral outcomes remains underexplored (Kolahchi et al., 2024). Microglia—the brain’s resident immune cells—are critical for synaptic regulation, and their morphology reflects shifts in functional state. Therefore, detailed morphological analysis offers critical insight into how microglia influence AD pathology. In this study, we performed a computational investigation quantitatively comparing hippocampal microglial morphology and cognitive performance in female and male 5xFAD mice, a transgenic AD model, and their wild-type controls. Leveraging digitally reconstructed cells from NeuroMorpho.org, we analyzed >6,000 microglial cells (AD n=4,834; Control n=1,176) using ten morphometrics; we further integrated 5-Choice Serial Reaction Time Task (5CSRTT) data from MouseBytes, an open-access rodent touchscreen database (Female: AD n=394, Control n=1289; Male: AD n=598, Control n=1028) to examine behavioral and morphological changes in 5xFAD mice during gender stratification. Except for the number of processes, all morphometrics differed significantly between AD and Control groups (Welch’s t-test, p < 0.0001). Principal component analysis (PCA) identified soma size and branching-based metrics as top contributors to PCA axes. Supervised models (random forest, gradient boosting; 5-fold CV) classified cells with 96% accuracy, while SHapley Additive exPlanations (SHAP) identified segments, soma radius, and segment length as key features in classification. Gender stratification revealed stronger AD-related effects in females (e.g. ~44% decrease in bifurcations and sections from control to AD), while males showed smaller changes across the same features. Analysis of 5CRSTT data revealed that males had larger differences in average threshold accuracy and average threshold condition, while females showed significantly larger differences in total perseverative correct responses (~34% decrease). ANOVA revealed that group (AD vs. Control) accounted for most variance in both morphology and behavioral metrics. Together, these results indicate that AD alters microglial morphology and cognitive performance in a sex-dependent manner, with females showing greater structural differences and both males and females exhibiting distinct behavioral deficits. These findings highlight the importance of considering sex in AD research and may inform the development of targeted, sex-specific interventions.

Avni is a high-school student researcher and advocate for the Alzheimer's Association with a strong interest in medicine and neuroscience. She was recently invited to present her research at the 2025 Society for Neuroscience Conference, the largest neuroscience conference in the world. Passionate about both STEM and the arts, Avni values the integration of creativity and scientific curiosity through STEAM. Her interdisciplinary approach drives her goal of pursuing a career in medicine while exploring projects that connect art and neuroscience to advance understanding of the human brain.