As Artificial Intelligence (AI) systems continue to grow in size and complexity, so does the difficulty of the quest for AI transparency. In a world of large models and complex AI systems, why do we explain AI and what should we explain? While explanations serve multiple functions, in the face of complexity humans have used and continue to use explanations to foster learning. In this position paper, we discuss how learning theories can be infused in the XAI lifecycle, as well as the key opportunities and challenges when adopting a learner-centered approach to assess, design and evaluate AI explanations. Building on past work, we argue that a learner-centered approach to Explainable AI (XAI) can enhance human agency and ease XAI risks mitigation, helping evolve the practice of human-centered XAI.

Background:Fatigue and chronic fatigue syndrome (CFS) have a considerable impact on quality of life, thus motivating people to develop skills for better management of their fatigue. While the number of commercial apps in this domain has increased, there has been limited exploration of their functionalities. Objective:This paper aims to address this research gap through a functionality review of 17 top-rated iOS and Android apps for fatigue, with the aim to articulate design implications for technologies focused on supporting the management of fatigue. Methods:We conducted a systematic search on the 2 most common app marketplaces, which resulted in the initial identification of 427 Apple apps and 1218 Google apps. From these, 17 apps were selected for review after applying a screening process to shortlist the top-rated apps. The functionalities of these apps were then coded through a week-long usage of each app for an expert evaluation leveraging authors' human-computer interaction (HCI) expertise. We looked for functionalities such as tracking and visualization seen in previous research on functionality reviews, in addition to interventional functionalities, which were informed by research on fatigue. Results:Findings reveal the prevalence of functionalities for tracking fatigue (8/17, 47%), related symptoms (8/17, 47%), for visualizing tracked content (10/17, 59%), for assessing the user's condition (2/17, 12%), and for providing interventions for the management of fatigue (12/17, 71%). Functionalities providing interventions for self-management of fatigue are surprisingly limited, with the most relevant ones including pacing (2/17, 12%) alongside energy estimation (2/17, 12%). Conclusions:The top-ranked apps for fatigue in the major marketplaces support 3 main functionalities under the scope of tracking fatigue along with related data, and visualizing such data, with limited provision of self-management interventions. Drawing from these findings, we articulate implications for the sensitive design of technologies to support the management of fatigue, including supporting hybrid tracking, combined visualizations to support sense-making of fatigue data with related factors, and supporting energy estimates and pacing interventions.

Engagement as a concept can explain why Digital Health Interventions (DHIs) produce individual variance in outcomes, and sometimes limited effectiveness, especially in practice. However, previous literature on engagement across different domains (e.g., Psychology, Implementation Science, Human-Computer Interaction) yields disparate conceptualizations, research methods, design strategies, and measurement methods. Therefore, this workshop aims to: bring together a diverse group of researchers within the field of DHIs with an interest in engagement; provide an overview of how engagement has been used, in terms of concept, measures, and strategies; work towards a shared understanding of how engagement, with its diverse measures and strategies, can be leveraged to inform the design, development, and evaluation of meaningful DHIs. We welcome submissions either as a description of a use case that includes: how engagement was defined, measured, designed for by our participants, as well as their lessons learned; or as a short position paper describing their interest in the topic, future plans for measuring/designing for engagement, and current challenges. Our post-workshop plans aim to draw from this transdisciplinary collaboration to document lessons learned on how to employ engagement in DHI development, research and design.

Although artificial intelligence (AI) shows growing promise for mental health care, current approaches to evaluating AI tools in this domain remain fragmented and poorly aligned with clinical practice, social context, and first-hand user experience. This paper argues for a rethinking of responsible evaluation – what is measured, by whom, and for what purpose – by introducing an interdisciplinary framework that integrates clinical soundness, social context, and equity, providing a structured basis for evaluation. Through an analysis of 135 recent *CL publications, we identify recurring limitations, including over-reliance on generic metrics that do not capture clinical validity, therapeutic appropriateness, or user experience, limited participation from mental health professionals, and insufficient attention to safety and equity. To address these gaps, we propose a taxonomy of AI mental health support types – assessment-, intervention-, and information synthesis-oriented – each with distinct risks and evaluative requirements, and illustrate its use through case studies.

As AI systems become increasingly integrated into our lives, the need to support appropriate human understanding of AI continues to grow. With new AI capabilities being deployed in different contexts, humancentered explainability is crucial to ensure people can interact with novel AI systems safely and effectively. To address evolving explainability needs, the field of Explainable AI (XAI) has produced numerous frameworks. But what do these frameworks entail and how can they be used in practice? What drives their development? As AI systems continue to grow in complexity, it is important to understand and reflect upon the value of these frameworks and their potential to address upcoming human-centered needs for XAI. Towards this, we performed a scoping review following the PRISMA-ScR procedure, gathering and analyzing a corpus of 73 papers to understand how XAI frameworks can support different stages of human-centered XAI design. We present a unified model and a set of guiding questions to help identify, compare and select relevant XAI frameworks across various design stages, making it easier for designers and researchers to apply human-centered approaches in real-world XAI contexts. We also analyze how frameworks are developed and evaluated, highlighting gaps and opportunities to improve both methodological as well as existing HCXAI practices.

Machine learning-augmented applications have the potential to be powerful tools for decision-making in healthcare. However, healthcare is a complex domain that presents many challenges. These challenges, such as medical errors, clinician-patient relationships, and treatment preferences, must be addressed to ensure fairness in ML-augmented healthcare applications. To better understand the influence these challenges have on fairness, 16 experienced engineers and designers with domain knowledge in healthcare technology were interviewed about how they would prioritise fairness in three healthcare scenarios (well-being improvement, chronic illness management, acute illness treatment). Using a template analysis, this work identifies the key considerations in the creation of fair ML for healthcare. These considerations clustered into categories related to technology, healthcare context, and user perspectives. To explore these categories, we propose the stakeholder fairness conceptual model. This framework aids designers and developers in understanding the complex considerations that stem from the building, management, and evaluation of ML-augmented healthcare applications, and how they affect the expectations of fairness. This work then discusses how this model may be applied when the health technology is directly provisioned to users, without a healthcare provider managing its use or adoption. This paper contributes to the understanding of fairness requirements in healthcare, including the effect of healthcare errors, clinician-application collaboration, and how the evaluation of healthcare technology becomes part of the fairness design process.

Smart home users often lack the technical expertise required to secure their devices and could benefit from the automated selection of security controls. In this paper, we explore the capabilities of inductive learning to adapt the requirements and system specification of a smart home system to identify security controls. We present preliminary results from using Inductive Learning via Answer Set Programming (ILASP) to learn how to produce (1) an updated system specification that enables benign behaviours while excluding malicious ones and (2) updated security requirements that the system should satisfy. We encode traces of benign and malicious execution traces from two smart home attack datasets (CICIoT2023 and IoT-23) into ILASP's language. ILASP could learn updated system specifications (to prevent DoS/Botnet attacks), new security requirements (to check for malware uploads and insecure protocols), and other integrity constraints that could be indicators of compromise. However, challenges remain when ILASP cannot perform the learning due to its sensitive syntax or complex system behaviour that lead to a large analysis space. Finally, we discuss how these limitations can be addressed in future work.

Conversational agents (CAs) are a tempting type of computer interface for assisting people's mental health due to their ability to simulate human-like interactions, however their integration within the broader social context of mental health management remains largely under-explored. Recognising that managing one's mental health is often a social rather than individual activity involving close persons such as partners, family, and friends, our research takes a social-orientation to mental health management. Utilising design cards that depict fictional, yet plausible CA concepts, we present the analysis of an interview study with 24 young adults to understand their views on CAs for both their own use and for a close person. Participants viewed CAs as potentially valuable complements to human support, but expressed concerns about over-reliance and replacement. Our analysis reveal key tensions, design considerations, and opportunities for integrating CAs into mental health ecosystems in ways that respect and enhance existing social support structures.

Security attacks are rising, as evidenced by the number of reported vulnerabilities. Among them, unknown attacks, including new variants of existing attacks, technical blind spots or previously undiscovered attacks, challenge enduring security. This is due to the limited number of techniques that diagnose these attacks and enable the selection of adequate security controls. In this paper, we propose an automated technique that detects and diagnoses unknown attacks by identifying the class of attack and the violated security requirements, enabling the selection of adequate security controls. Our technique combines anomaly detection to detect unknown attacks with abductive reasoning to diagnose them. We first model the behaviour of the smart home and its requirements as a logic program in Answer Set Programming (ASP). We then apply Z-Score thresholding to the anomaly scores of an Isolation Forest trained using unlabeled data to simulate unknown attack scenarios. Finally, we encode the network anomaly in the logic program and perform abduction by refutation to identify the class of attack and the security requirements that this anomaly may violate. We demonstrate our technique using a smart home scenario, where we detect and diagnose anomalies in network traffic. We evaluate the precision, recall and F1-score of the anomaly detector and the diagnosis technique against 18 attacks from the ground truth labels provided by two datasets, CICIoT2023 and IoT-23. Our experiments show that the anomaly detector effectively identifies anomalies when the network traces are strong indicators of an attack. When provided with sufficient contextual data, the diagnosis logic effectively identifies true anomalies, and reduces the number of false positives reported by anomaly detectors. Finally, we discuss how our technique can support the selection of adequate security controls.

In this column, we illustrate real-world scenarios in which modern systems cannot preserve security during operation. We examine the notion of sustainable security and discuss the challenges to engineering sustainably secure systems.

Self-monitoring of mood and lifestyle habits is the cornerstone of many therapies, but it is still hindered by persistent issues including inaccurate records, gaps in the monitoring, patient burden, and perceived stigma. Smartwatches have the potential to deliver enhanced self-reports, but their acceptance in clinical mental health settings is unexplored and rendered difficult by a complex theoretical landscape and need for a longitudinal perspective. We present the Mood Monitor smartwatch application for mood and lifestyle habits self-monitoring. We investigated patient acceptance of the app within a routine 8-week digital therapy. We recruited 35 patients of the UK’s National Health Service and evaluated their acceptance through three online questionnaires and a post-study interview. We assessed the clinical feasibility of the Mood Monitor by comparing clinical, usage, and acceptance metrics obtained from the 35 patients with a smartwatch with those from an additional 34 patients without a smartwatch (digital treatment as usual). Findings showed that the smartwatch app was highly accepted by patients, revealed which factors facilitated and impeded this acceptance, and supported clinical feasibility. We provide guidelines for the design of self-monitoring on a smartwatch and reflect on the conduct of human-computer interaction research evaluating user acceptance of mental health technologies.

Politeness is important in human-human interaction when asking people to engage in sensitive conversations. If politeness manifests similarly in human-chatbot interaction, it may play an important role in the design of sensitive chatbot interactions such as those for providing mental health support. Our mixed methods study (N = 39) contributes findings on how the use of politeness by chatbots, for the mental healthcare activity of mood logging, is perceived by users. Our study combined a within-participants controlled experiment, whereby participants interacted with three prototype chatbots differing in their use of politeness, with semi-structured interviews. Our analysis demonstrates that a chatbot’s use of politeness can impact how a participant experiences interacting with it, both positively and negatively. While politeness can be experienced as caring, supportive, and encouraging, it can also be experienced as overly apologetic, condescending, and untrustworthy. We discuss the nuances of using politeness in conversational interaction design, setting out a research agenda for polite conversational interaction.

Digital mental health interventions (DMHIs) have potential to provide effective and accessible care to entire populations, but low client uptake and engagement are significant problems. Few prior studies explore the lived experiences of non-engagers, because reaching this population is inherently difficult. We present an observational inquiry into the barriers to sign-up and early use of a DMHI, along with reasons for initial interest in the DMHI. We collected 205 online questionnaire responses and 20 interviews from self-referring participants across four healthcare ecosystems in the UK and US. Questionnaire results revealed that uncertainty about DMHI usefulness and usability were the main barriers to uptake, whereas forgetting about it, not finding time for it and not finding it useful were the main barriers to early engagement. Participants reported multiple reasons for considering the DMHI, reflecting the contextual, subjective nature of mental health. Our thematic analysis generated themes around 1) the need for human connection, 2) the impact of self-stigma on help-seeking, 3) the lack of knowledge around DMHIs and psychological therapy, 4) the desire for personally relevant care, and 5) the fluctuating, perennial nature of mental health. We discuss implications for DMHI design, implementation and future research, as well as transdisciplinary opportunities.

With software systems permeating our lives, we are entitled to expect that such systems are secure-by-design, and that such security endures throughout the use of these systems and their subsequent evolution. During my PhD, I aim to engineer sustainable adaptive security solutions that reflect such enduring protection in the dynamically changing security theatre of cyber-physical systems. I have chosen the example of a smart home as a cyber-physical system to motivate & illustrate sustainable adaptive security, discuss challenges for sustainably secure systems, and my research plan for engineering them.

Mood logging, where people track mood-related data, is commonly used to support mental healthcare. Speech agents could prove benefcial in supporting mood logging for clients. Yet we know little about how Mental Healthcare Practitioners (MHPs) view speech as a tool to support current care practices. Through a thematic analysis of semi-structured interviews with 15 MHPs, we show that MHPs see opportunities in the convenience, and the data richness that speech agents could aford. However, MHPs also saw this richness as noisy, with using speech potentially diminishing a client's focus on mood logging as an activity. MHPs were wary of overusing AI-based tools, expressing concerns around data ownership, access and privacy. We discuss the role of speech agents within blended care, outlining key considerations when using speech for mood logging in a blended mental healthcare context.

Anxiety disorders are the most common mental health problem, and cognitive-behavioral therapy is one of the most widely used, evidence-based treatments. While several mobile apps for anxiety that integrate cognitive-behavioral therapy (CBT) techniques exist, ma- jor challenges remain concerning uptake and engagement. Personalization is one strategy that can be used to improve client engagement, and integrating therapist input is one mechanism for such personalization. This study aims to understand therapist practices and identify new possibilities for delivering intervention content between face-to-face CBT therapy sessions. It comprised semi-structured interviews, followed by a series of ideation activities, and thematic analysis of the data. The results showed the central role of clients in shaping the content of therapy sessions, their challenges with homework practice, and therapists’ diverse practices. Analysis of the ideation activities elaborated the potential role of therapists in the personalization of apps for anxiety. We conclude with takeaways for designers of personalized mental health mobile applications.

The use of smartphone-based Serious Games in mental health care is an emerging and promising research field. Combining the intrinsic characteristics of games (e.g., interactiveness, immersiveness, playfulness, user-tailoring and engaging nature) with the capabilities of smartphones (e.g., versatility, ubiquitous connectivity, built-in sensors and anywhere–anytime nature) yields great potential to deliver innovative psychological treatments, which are engaging, effective, fun and always available. This article presents a scoping review, based on the PRISMA (scoping review extension) guidelines, of the field of smartphone-based serious games for mental health care. The review combines an analysis of the technical characteristics, including game design, smartphone and game-specific features, with psychological dimensions, including type and purpose of use, underlying psychological frameworks and strategies. It also explores the integration of psychological features into Serious Games and summarizes the findings of evaluations performed. A systematic search identified 40 smartphone-based Serious Games for mental health care. The majority consist of standalone and self-administrable interventions, applying a myriad of psychological strategies to address a wide range of psychological symptoms and disorders. The findings explore the potential of Serious Games as treatments and for enhancing patient engagement; we conclude by proposing several avenues for future research in order to identify best practices and success factors.

Music-based reminiscence has the potential to positively impact the psychological well-being of older adults. However, the aging process and physiological changes, such as memory decline and limited verbal communication, may impede the ability of older adults to recall their memories and life experiences. Given the advanced capabilities of generative artificial intelligence (AI) systems, such as generated conversations and images, and their potential to facilitate the reminiscing process, this study aims to explore the design of generative AI to support music-based reminiscence in older adults. This study follows a user-centered design approach incorporating various stages, including detailed interviews with two social workers and two design workshops (involving ten older adults). Our work contributes to an in-depth understanding of older adults' attitudes toward utilizing generative AI for supporting music-based reminiscence and identifies concrete design considerations for the future design of generative AI to enhance the reminiscence experience of older adults.

In treating depression and anxiety, just over half of all clients respond. Monitoring and obtaining early client feedback can allow for rapidly adapted treatment delivery and improve outcomes. This study seeks to develop a state-of-the-art deep-learning framework for predicting clinical outcomes in internet-delivered Cognitive Behavioural Therapy (iCBT) by leveraging large-scale, high-dimensional time-series data of client-reported mental health symptoms and platform interaction data. We use de-identified data from 45,876 clients on SilverCloud Health, a digital platform for the psychological treatment of depression and anxiety. We train deep recurrent neural network (RNN) models to predict whether a client will show reliable improvement by the end of treatment using clinical measures, interaction data with the iCBT program, or both. Outcomes are based on total improvement in symptoms of depression (Patient Health Questionnaire-9, PHQ-9) and anxiety (Generalized Anxiety Disorder-7, GAD-7), as reported within the iCBT program. Using internal and external datasets, we compare the proposed models against several benchmarks and rigorously evaluate them according to their predictive accuracy, sensitivity, specificity and AUROC over treatment. Our proposed RNN models consistently predict reliable improvement in PHQ-9 and GAD-7, using past clinical measures alone, with above 87% accuracy and 0.89 AUROC after three or more review periods, outperforming all benchmark models. Additional evaluations demonstrate the robustness of the achieved models across (i) different health services; (ii) geographic locations; (iii) iCBT programs, and (iv) client severity subgroups. Results demonstrate the robust performance of dynamic prediction models that can yield clinically helpful prognostic information ready for implementation within iCBT systems to support timely decision-making and treatment adjustments by iCBT clinical supporters towards improved client outcomes.