Rethinking ADHD: Lifespan, Sex, and Hormonal Insights through Computational Modeling for the Common Good

Background and Purpose

ADHD in girls and women is often missed; not because it is absent, but because our models were never designed to see it. Attention-Deficit/Hyperactivity Disorder (ADHD) affects approximately 3–5% of children globally and often persists into adulthood. Significant gender disparities remain in diagnosis, treatment, and outcomes. Boys are more likely to exhibit hyperactivity and externalizing behaviors aligned with traditional diagnostic criteria, whereas girls and women more commonly present with inattention, emotional dysregulation, and internalizing symptoms such as anxiety and depression. Hormonal fluctuations across the lifespan (e.g., puberty, menstrual cycles, pregnancy, menopause) further shape symptom expression, contributing to systematic underrecognition and delayed diagnosis among females. Guided by a health and gender equity framework and aligned with SSWR 2027’s theme of advancing rigorous inquiry for the common good, this study aims to (1) identify sex-specific neurobiological patterns associated with ADHD, (2) develop predictive models that improve diagnostic inclusivity, and (3) translate findings into equitable, family-centered screening approaches.

Methods

We analyzed functional MRI (fMRI) connectome data from 1,213 participants (~19,900 features) using machine learning models to predict ADHD diagnosis. Feature engineering focused on brain connectivity across key networks, including the executive control network (attention regulation and working memory), salience network (attention switching), and limbic system (emotional regulation). Analyses examined sex-based differences in neural connectivity patterns using a secondary dataset (WiDS Datathon ADHD dataset) and supervised learning approaches.

Results

Altered connectivity in the executive control network was associated with deficits in attentional regulation, while salience network disruptions were linked to impaired attention switching. Notably, limbic system connectivity emerged as a salient predictor of ADHD, particularly in relation to emotional dysregulation; an area more commonly observed in girls and women but underrepresented in existing diagnostic frameworks. These findings suggest that current models may insufficiently capture neurobiological patterns relevant to female ADHD presentations, contributing to inequities in identification and care.

Conclusions and Implications

This study advances gender equity in mental health by demonstrating how computational modeling can uncover overlooked ADHD patterns in girls and women. A prototype digital tool (“MomBud”) illustrates the potential for family-centered early screening and intervention. Implications for social work include integrating equity-informed, data-driven tools to support earlier and more accurate identification of ADHD in underserved populations.

Future research will expand this work by engaging University of California campuses; including UC Berkeley, UC Davis Health, and UCSF, to conduct surveys with college student populations, with particular attention to gender differences in ADHD experiences. In addition, future studies will incorporate multimodal data, including biomarkers to detect emotional states and track mood patterns over time, alongside hormonal, environmental (e.g., adverse childhood experiences), and behavioral signals across the lifespan. These efforts will support the development of more inclusive, longitudinal, and ecologically valid models while reducing bias, improving generalizability, and ensuring culturally responsive, community-engaged design.