UXR PoV for Neuroinclusive Emotion Regulation

Technical Summary: UXR PoV for Neuroinclusive Emotion Regulation

Problem Statement
Attention-Deficit/Hyperactivity Disorder (ADHD) is characterized by inattentiveness, hyperactivity, and impulsivity, but a critical yet often under-addressed component is difficulty with emotional regulation (ER). These difficulties lead to emotional instability, reduced quality of life, and increased vulnerability to comorbid mental health conditions. While Digital Health Interventions (DHIs) offer accessible support for ER, current systems frequently fail to address the specific cognitive and emotional variability inherent in ADHD. Existing design methodologies often lack the theoretical rigor to translate psychological models into concrete, testable design outcomes and insufficiently accommodate neurodiversity. Furthermore, there is a gap in structured User Experience Research (UXR) methodologies that effectively bridge psychological insight with design practice, particularly when leveraging emerging Generative AI (GenAI) tools.

Methodology
The study employs an AI-augmented systematic literature review (SLR) integrated with the UXR Point-of-View (PoV) framework. The methodology follows a four-stage process, adapted from the "Developing an AI-Powered UX Research Point of View" workshop, to translate empirical evidence into actionable design guidance:

1. Leveraging GenAI and the UXR PoV Framework: Researchers conducted a systematic literature search (2020–2025) across ACM Digital Library, PubMed, and Scopus, resulting in 42 empirical studies on ADHD and ER. This dataset, along with UXR PoV framework documents, was uploaded to four GenAI systems (ChatGPT v5.2, NotebookLM, Gemini v3, and Microsoft Copilot). The AI was prompted to identify thematic patterns, cluster intervention features, and generate testable hypotheses linking UX variables (e.g., feedback, autonomy) to ER outcomes.
2. Establishing a Foundational Plan and Roadmap: GenAI was used to simulate theoretical integration and stakeholder mapping. This stage contextualized ER within the lived experiences of adults with ADHD, aligning design goals with three psychological frameworks: Dialectical Behaviour Therapy (DBT), Self-Determination Theory (SDT), and the COM-B model. It also mapped the roles, needs, and challenges of five key stakeholder groups: ADHD users, clinicians, UX designers, researchers, and developers.
3. Insight Generation and Play Card Development: Using the "Building Blocks" model (Foundation, Data Collection, Insight Generation, PoV), the team synthesized empirical data and psychological theory into ten "UXR Play Cards." GenAI assisted in clustering data into themes (e.g., attention variability, feedback sensitivity) and linking these to design principles. Each card addresses a specific UX challenge, providing a title, illustrative quote, issue type, best practice, and required UXR skills.
4. Crafting PoV Narratives: The final stage involved using GenAI to refine the Play Card insights into coherent, evidence-based Point-of-View narratives tailored to specific stakeholders. The AI helped adjust tone and structure to ensure clinical accuracy and user experience clarity while maintaining alignment with neurodiversity and therapeutic language.

Throughout the process, human researchers maintained oversight, using GenAI as a "co-analytic tool" for synthesis and hypothesis formation rather than a substitute for human interpretation. Comparative analysis across the four AI models was conducted to assess interpretive depth and theoretical integration.

Key Contributions and Results
The primary contribution of this work is a replicable, bias-aware framework for integrating Generative AI into UXR practice to support neuroinclusive digital mental health design. The specific outputs include:

• Ten Theory-Informed Hypotheses: The study generated ten hypotheses linking design features to ER outcomes, such as the need for simplified emotional cues to reduce cognitive overload (H1), consistent AI tone to build trust (H2), and customizable goal systems to enhance autonomy (H3).
• Stakeholder Matrix: A structured mapping of five stakeholder groups, clarifying their distinct needs (e.g., clinicians requiring data visibility; designers needing neuroinclusive heuristics) and the systemic tensions between them (e.g., balancing emotional safety with technical feasibility).
• Ten UXR Play Cards: A tangible set of design artifacts that translate psychological mechanisms (DBT, SDT, COM-B) and empirical findings into actionable guidance. These cards cover critical areas such as reducing cognitive load, ensuring emotional predictability in AI, empowering user agency, designing for emotional safety, and creating inclusive AI emotion models.
• Stakeholder-Specific PoV Narratives: Four cross-cutting design pillars—empathy, predictability, autonomy, and accessibility—were identified to unify the design approach across different stakeholder perspectives.

Significance and Claims
The paper claims that this methodology advances human-centred and neuroinclusive approaches to digital mental health by formalizing the translation of raw data into actionable, empirically testable design decisions. It posits that GenAI can effectively accelerate data synthesis and hypothesis generation but must be paired with human-led reflexive validation to mitigate interpretive risks and biases inherent in large language models.

The authors emphasize that while GenAI demonstrated value in scaling research synthesis, it lacked the capacity to interpret affective nuance or empathetically engage with neurodiverse lived experiences. Therefore, the proposed framework positions the researcher as a "curator" who reconciles AI-generated outputs with contextual knowledge and psychological theory. The study concludes that this hybrid, bias-aware approach offers a practical mechanism for addressing the weak theoretical integration often found in current DHIs, providing a structured pipeline from evidence to design for adults with ADHD. Future work is proposed to focus on empirical validation of these designs, cross-diagnostic generalization, and the development of open-access toolkits for neuroinclusive design.