Exploring Late-Onset ADHD in Women: A Data-Driven Approach Using AI and Socio-Demographic Analysis

Abstract

Due to internalized symptoms and social masking, women are often diagnosed with ADHD later in life, making gender differences in diagnosis a significant and understudied topic. AI is being increasingly used to enhance diagnostic tools and uncover intricate patterns in behavioral, neuroimaging, and medical record data, although the majority of applications do not stratify data by gender. This disparity could unintentionally strengthen preexisting biases in conventional diagnostic systems. To determine whether and how AI-based approaches to ADHD diagnosis consider gender differences, we conducted a systematic review of the literature in this study. We examine peer-reviewed studies that utilize AI methods across various data modalities between 2013 and 2025. Studies are categorized by publication time, gender focus, AI methodology, and data type. Our analysis reveals that, although AI has the potential to enhance diagnostic precision, the lack of gender-specific evaluation remains a significant drawback. We contend that to guarantee more equitable mental health results, particularly for historically underdiagnosed populations like women with ADHD, future AI systems should specifically include gender-sensitive design and demographic fairness.