AI and Gamification for Employees with ADHD: A Tool for Enhancing Workplace Focus & Productivity
Tahani Jaser Alahmadi, Ruyouf AlKhuzaim, Meral AlHelwah, Abbrar AlOtaibi, Dima AlTammami, Daad AlSaqer · 2025 · Proceedings of the 22nd International Web for All Conference (W4A) · doi:10.1145/3744257.3744260
Summary
This paper presents WADAQ, an AI-driven gamification tool designed to support employees with ADHD in improving focus, task management, and workplace productivity. The system operates in three stages: first, machine learning classification models predict whether a user is likely to have ADHD using data from the 2022 National Survey of Children's Health (54,104 participants, 490 features reduced to 15 via chi-square feature selection). Users who screen positive then complete a digital version of the Vanderbilt ADHD Diagnostic Parent Rating Scale to determine their ADHD subtype—inattentive, hyperactive/impulsive, or combined. Based on the subtype classification, WADAQ recommends tailored mini-games using reinforcement learning (Q-learning) to dynamically adjust difficulty. Four games were selected after reviewing existing ADHD interventions: the Fin game and balloon analogue risk task for impulsivity/hyperactivity, and Falling Dots and Through for inattention. Each targets specific cognitive skills like impulse control, decision-making, sustained attention, and task management. The tool includes a real-time performance dashboard tracking metrics such as correctness rate, error rate, average time, and best time, giving both employees and employers visibility into progress. The authors frame WADAQ not just as a therapeutic tool but as a workplace inclusion instrument that could inform hiring practices, accommodations, and organizational ADHD policy.
Key findings
Among the four classification models tested (Decision Tree, Random Forest, SVM, K-Nearest Neighbours), SVM achieved the highest accuracy at 97% after hyperparameter optimization with GridSearchCV—a 12 percentage point improvement over its baseline hold-out accuracy of 85%. Random Forest and Decision Tree both achieved 94%, while KNN reached 92%. The tool's accessibility design follows multiple WCAG 2.1 criteria: limited colour palettes to reduce cognitive overload (1.4.1), optimized contrast (1.4.3, 1.4.11), adjustable timers for varying attention spans (2.2.1), structured navigation with headings and labels (2.4.6), clear instructions with help buttons (3.3.2), logical content sequencing (1.3.2), and Arabic/English language toggle (3.1.1). The reinforcement learning component uses Q-learning to adapt game difficulty in real time based on user performance, creating a personalized training experience. The authors note that while existing serious games for ADHD like Plan-It Commander and Focus have shown promise, they are limited by generalization issues and outdated datasets—problems WADAQ aims to address with more current data and individualized strategies.
Relevance
This paper sits at an interesting intersection of AI, gamification, and workplace accessibility for neurodivergent employees. For accessibility practitioners, the most valuable contribution is the explicit mapping of WCAG 2.1 success criteria to the needs of users with ADHD—demonstrating how existing web accessibility standards can be applied to support cognitive accessibility in practical tool design. The concept of integrating ADHD assessment and cognitive training into workplace systems raises important questions about privacy, disclosure, and the ethics of employer access to neurodevelopmental data. The tool is still in early stages—no real-world user testing with employees has been conducted yet, and the training data comes from a children's health survey applied to adult contexts. However, the framework of combining ML-based screening with adaptive, gamified cognitive training represents a promising direction for digital workplace accommodations that go beyond static adjustments.
Tags: ADHD · gamification · machine learning · reinforcement learning · workplace accessibility · cognitive accessibility · digital therapeutics · employment
Standards referenced: WCAG 2.1 · DSM-IV