Artificial intelligence (AI) Large Language Models (LLMs) are increasingly being used in mental health training and therapeutic interventions. This narrative review explores how LLMs can support simulation-based training for mental health practitioners by simulating patient interactions. Key topics include the use of co-production techniques, cost considerations, and the role of Retrieval-Augmented Generation (RAG) in enhancing training accuracy. The review highlights the importance of evaluation methods and expert-guided enhancements. Although LLMs offer valuable training opportunities, they should complement, rather than replace traditional face-to-face training methods. Collaboration across disciplines is essential in refining AI chatbots and enhancing their effectiveness in therapeutic training. In conclusion, the integration of AI in mental health practitioner training shows promise for addressing the growing demands within mental health care. However further research is required to develop standardized approaches and improve the realism of AI simulations.

This editorial introduces a special issue in digital mental health and wellbeing and includes topics such as digital wellbeing, chatbots, virtual reality and youth mental health. This editorial also discusses the potential adverse psychological effects of an artificial intelligence (AI) centric future. We emphasise the need to be ‘wellbeing centric’ and not technology centric. AI can expedite tasks and ‘think’ on our behalf, but if the goal of humanity is to promote ‘wellbeing’ then ‘wellbeing’ should be a core design principle and an evaluation metric to assess AI technologies. We discuss how the potential adverse effects of social media can be a good case study for how we should assess the AI future – especially in the era of generative AI and AI companions. To illustrate a point, we explore a fictional scenario (akin to Laplace’s demon) that includes an AI agent that makes minute-by-minute ‘life recommendations’. We explore how such an AI could negatively affect our mental wellbeing, agency, autonomy and sense of free will – even if the ‘AI always knows best’. We explore conscious AI and the potential impact of AI from a positive psychology perspective.

Employers now acknowledge the crucial role of employee wellbeing in promoting productivity, positive relationships, and engagement, as well as its impact on absenteeism and presenteeism. Consequently, there is an increasing need for affordable, evidence-supported, scalable innovative approaches to improve employee wellness. This paper presents usage analysis of a digital employee wellbeing platform, created by Inspire, a mental health social enterprise. The platform has several self-help components, including a chatbot that delivers mental health self-assessments, CBT -based e-learning modules, and a mood tracker. Analysis was conducted using the machine learning technique k-means clustering and descriptive analytics. Through the analysis of user tenure (i.e. the time interval between the first and last day of a user who engaged with the platform), total interactions, daily interactions, and number of unique days on the platform, K-means clustering successfully identified three user groups: short-term (95.5% of users), intermediate (3.4% of users), and long-term users (1.1 % of users). By using these analysis techniques, we can understand how employees utilize a digital employee wellbeing platform, helping to design more effective and personalized solutions.

As the proliferation of video content continues, and many video archives lack suitable metadata, therefore, video retrieval, particularly through example-based search, has become increasingly crucial. Existing metadata often fails to meet the needs of specific types of searches, especially when videos contain elements from different modalities, such as visual and audio. Consequently, developing video retrieval methods that can handle multi-modal content is essential. In designing our novel video retrieval framework named Multi-modal Video Search by Examples (MVSE)1, we focused on accuracy (precision and recall), efficiency (retrieval time in seconds), interactivity, and extensibility, with key components including advanced data processing and a user-friendly interface aimed at enhancing search effectiveness and user experience. With the advent of Large Language Models (LLMs), the interaction between multimodal data, including image and audio has been transformed with a significant leap forward towards a bigger goal of artificial general intelligence. This workshop aims to bring together experts from diverse domains to explore the possibilities of developing novel ways of multimodal data search, understanding and interaction.

The development and monitoring of Photovoltaic (PV) installations is of great interests for the Chinese energy management agency in recent years. The traditional land change detection of PV installations has issues pertaining to low efficiency and high missed detection rates. Therefore, this paper explores an efficient and high accurate detection method of PV installations land using changes from remote sensing images in order to help relevant stakeholders to better manage and monitor urban energy and environment. In this paper, Full Convolutional Network (FCN) and classical segmentation convolutional network (U-Net) based deep learning algorithms are used to build change detection models. To evaluate the model performance, we have built the change detection dataset from Northeast Petroleum University - Photovoltaic Remote Sensing Dataset (NEPU-PRSD) of PV installations in Western China. The experimental results show that both models can achieve good accuracy in change detection regarding PV installations.

Abstract Technological advances have facilitated new approaches to support different needs in healthcare environments. In particular, in the field of rehabilitation therapies, we can find software applications that have been developed to support the performance of specific exercises, often with different customization options. However, there are still many gaps that need to be addressed to provide more and better global solutions. In this article, we present a novel system aimed at physiotherapists, which allows them to create new rehabilitation exercises designed to meet the specific needs of their patients. This implies the creation of individualized therapies that contribute to a better and faster recovery of patients. The system consists of a virtual 3D environment with a 3D skeleton representing the patient. The physiotherapist can interact with the skeleton to design the desired postures that the patient should practice according to their personal limitations. The system allows physiotherapists to compose a complete and personalized set of exercises. Alternatively, the physiotherapist may also create the postures by means of a motion sensing device using motion- and voice-based interaction. Both options pose research challenges that the authors have addressed to provide the solution presented in this paper.

This study reports on the development and ‘in the wild’ trialling of a chatbot (ChatPal) which promotes good mental wellbeing. A stakeholder-centered approach for design was adopted where end users, mental health professionals and service users were involved in the design which was centered around positive psychology. In the wild usage of the chatbot was investigated from Jul-20-Mar-21. Exploratory analyses of usage metrics were carried out using the event log data. User tenure, unique usage days, total chatbot interactions and average daily interactions were used in K-means clustering to identify user archetypes. The chatbot was used by a variety of age groups (18-65+) and genders, mainly those living in Ireland. K-means clustering identified three clusters: sporadic users (n=4), frequent transient users (n=38) and abandoning users (n=169) each with distinct usage characteristics. This study highlights the importance of event log data analysis for making improvements to the mental health chatbot.

In this paper, we use the HappyDB (which is a corpus of more than 100,000 happy moments or happiness statements) to train machine learning classifiers to classify the type of happiness statements, i.e., whether they are related to different categories, for example Achievement or Affection. Having identified the best performing classifier, we then sought to assess if the classifier had variable performance when tested using happiness statements from different demographic groups, such as those written by a married or single person, female or male, young or old and whether they are a parent or non-parent. Three different classifiers were initially used in this classification task, to determine classification accuracy. Having determined the best performing model (the convolutional neural network - CNN, deep learning algorithm), this model was then used for further analysis of results per cross sectional demographic groups. The CNN achieved an F1 score of 0.897 but had variable performance when tested on different demographic groups. Generally, we found that accuracy of prediction within this dataset declines with age, where the results for certain sub-groups were declining with increased age or flatlining, except for the single parents' sub-group. This may be due to decreased numbers in these particular sub-groups, where the algorithm did not learn the patterns in the happiness statements for this cohort, due to a sparsity of training data for the sub-group. Results show that there is likely a change in word patterns in happiness statements for different demographics.

The objective of this study was to understand the attitudes of professionals who work in mental health regarding the use of conversational user interfaces, or chatbots, to support people’s mental health and wellbeing. This study involves an online survey to measure the awareness and attitudes of mental healthcare professionals and experts. The findings from this survey show that more than half of the participants in the survey agreed that there are benefits associated with mental healthcare chatbots (65%, p < 0.01). The perceived importance of chatbots was also relatively high (74%, p < 0.01), with more than three-quarters (79%, p < 0.01) of respondents agreeing that mental healthcare chatbots could help their clients better manage their own health, yet chatbots are overwhelmingly perceived as not adequately understanding or displaying human emotion (86%, p < 0.01). Even though the level of personal experience with chatbots among professionals and experts in mental health has been quite low, this study shows that where they have been used, the experience has been mostly satisfactory. This study has found that as years of experience increased, there was a corresponding increase in the belief that healthcare chatbots could help clients better manage their own mental health.

Abstract The discovery of new medications in a cost-effective manner has become the top priority for many pharmaceutical companies. Despite decades of innovation, many of their processes arguably remain relatively inefficient. One such process is the prediction of biological activity. This paper describes a new deep learning model, capable of conducting a preliminary screening of chemical compounds in-silico. The model has been constructed using a variation autoencoder to generate chemical compound fingerprints, which have been used to create a regression model to predict their LogD property and a classification model to predict binding in selected assays from the ChEMBL dataset. The conducted experiments demonstrate accurate prediction of the properties of chemical compounds only using structural definitions and also provide several opportunities to improve upon this model in the future.

This study uses classifications of Metropolitan and Non-Metropolitan behavioural archetypes of General Practitioner practice in Northern Ireland. Of the 333 practices operating in Northern Ireland at the start of the study period (March 2018), 90 were classified as Metropolitan and 243 as Non-Metropolitan. This paper seeks to examine any associations between deprivation and each archetype and to investigate what their prescription behaviours would be when controlled for deprivation. It was found that for each archetype, as the deprivation level of the area in which a practice was located increased, so too did the prescribing levels associated with that practice. A large proportion of Metropolitan practices (52.2%) were located in high deprivation areas and these practices had prescribing levels that were 40.3% higher in the number of items prescribed per registered patient than practices in low deprivation areas within the same archetype. Only 13.9% of Non-Metropolitan practices were located in high deprivation areas and had levels of prescribing 11.2% greater than practices in areas of low deprivation within the same archetype. A comparison of only practices in low deprivation areas in both archetypes found that higher prescribing levels were seen in Non-Metropolitan practices, the complete opposite of the trend observed when comparing all practices where Metropolitan practices show higher prescribing levels.

This study uses classifications of Metropolitan and Non-Metropolitan behavioural archetypes of General Practitioner practice in Northern Ireland and seeks to examine any associations between practice size and each archetype. It was found that the highest prescribing levels were in Small practices with two registered doctors for both archetypes. The lowest levels of prescribing were found in Single-Handed practices with only one registered doctor in Non-Metropolitan areas whilst Large practices with five or more registered doctors had the lowest prescribing levels in Metropolitan areas. One possible reason for Large practices having the lowest prescribing rates in Metropolitan areas may be the availability of more alternatives to medication. The highest prescribing levels in Non-Metropolitan areas were almost 4% higher than the lowest. This difference rose to 32% in Metropolitan areas. Further research into this difference is needed with deprivation levels found in Metropolitan areas being a possible factor. Examining each practice size individually showed that the same archetypes (i.e. Metropolitan and Non-Metropolitan) is also observed at this level with the number of registered patients increasing in line with the size of the practice although practices in Non-Metropolitan areas generally had more patients than those in Metropolitan areas.

This research presents a novel aggregating method for constructing an aggregated topic model that is composed of the topics with greater coherence than individual models. When generating a topic model, a number of parameters have to be specified. The resulting topics can be very general or very specific, which depend on the chosen parameters. In this study we investigate the process of aggregating multiple topic models generated using different parameters with a focus on whether combining the general and specific topics is able to increase topic coherence. We employ cosine similarity and Jensen-Shannon divergence to compute the similarity among topics and combine them into an aggregated model when their similarity scores exceed a predefined threshold. The model is evaluated against the standard topics models generated by the latent Dirichlet allocation and Non-negative Matrix Factorisation. Specifically we use the coherence of topics to compare the individual models that create aggregated models against those of the aggregated model and models generated by Non-negative Matrix Factorisation, respectively. The results demonstrate that the aggregated model outperforms those topic models at a statistically significant level in terms of topic coherence over an external corpus. We also make use of the aggregated topic model on social media data to validate the method in a realistic scenario and find that again it outperforms individual topic models.

The MIDAS consortium is a partnership involving health authorities, and technical big data experts from universities, research institutions, MNCs and SMEs across six EU countries (UK (NI and England), Ireland, Belgium, Finland, Spain and Slovenia), and the USA. The management of big data for 'health in all' poses a significant challenge for health policy makers. This challenge is addressed by the MIDAS project through the development of an integrated solution enabling knowledge liberation from data silos and unification of heterogeneous big data sources that can provide evidence-based actionable information and transform the way care is provided.

Smartphone-based digital phenotyping can provide insight into mood, cognition and behaviour. In this study, data analytics was carried out with data generated from a maternal mental health app to address the following question: what is the temporal behaviour of users when completing ecological momentary assessments (EMAs) with EMAs in the form of mental health scales versus EMAs in the form of mood logs? The methodology involved using the Health Interaction Log Data Analytics (HILDA) pipeline to analyse 1461 app users. Clustering was used to characterise archetypical user engagement with the two forms of EMA. Users preferred mood log EMAs, with 6993 mood log completions compared to 2129 scale completions. Users are more willing to log moods at 9am and 12pm and complete mental health scales between 8pm and 10pm. The fewest number of mood logs and scale completions take place on Saturday followed by a Sunday. Whilst ‘happiness’ is the dominant mood during day times, ‘anxiety’ and ‘sadness’ peak during night times. The overall findings are that users prefer completing mood log EMAs and that the temporal behaviour of users engaging with EMAs in the form of mental health scales are distinctly different from how they engage with mood logs.

Welcome to this special issue ‘Highlights from the European Conference on Cognitive Ergonomics (ECCE-2019)’ of Behaviour & Information Technology, presenting six papers selected from the programme ...

The 32nd International BCS Human Computer Interaction Conference was held in Belfast, Northern Ireland in July 2018. The conference proceedings included over 230 papers with various topics and applications in Human Computer Interaction (HCI). There were over 60 full papers, more than 90 work-in-progress papers, almost 30 position papers, 17 industry talks (2 submitted papers), around 25 workshop papers, 17 doctoral consortium papers and 4 interactions gallery papers. Pre-conference activities included a doctoral consortium, a masterclass in chatbot design and five workshops. This introduction presents an analysis of the topics presented at the 2018 conference providing a reflection of the popular themes that exist in HCI as a research discipline. Interestingly, the doctoral consortium papers included topics such as affective computing, tangible user interfaces, information quality, behaviour change interventions in digital healthcare, playful interactions, storytelling, trust, human–data interaction, reading behaviour, interactive public displays, haptic devices, visualizing personal health data and decision-making. The five workshops included the topics: affective computing; using data to design user interfaces; human-centred design in intelligent environments; digital technology for older people; and digital health. Perhaps as expected, the two application areas include interactive technologies for older people and digital health which is well aligned to the current public need and growing demands.

We are living in an ageing population with an escalation in chronic illnesses including dementia and other age related diseases. People living with dementia often continue to live at home and are supported by caregivers and next of kin. It is often important to monitor the wellbeing of people living with dementia in order to measure their level of independence and to provide proper support at the time of need as well as supporting their quality of life. Some researchers have focused on monitoring physical wellbeing and activities of daily living (ADL). However, there has been a paucity of research focussed on monitoring mood, affect and the emotional wellbeing of people living with dementia, despite these people experiencing frustration, agitation, depression and social isolation to name but a few known effects. As a result, the SenseCare project aims to build an affective computing platform that uses sensors placed in the home environment to monitor moods, affect and the emotional wellbeing of people living with dementia. This platform is being iteratively designed and will likely use plug-n-play sensors such as passive infrared, wearables and camera technologies to infer emotions from facial expressions, voice intonations and physical behaviour and other modalities. However, it is important to interact iteratively with people living with dementia and their caregivers in order to understand their profound needs. In this study, we report on two focus groups that were conducted to elicit user stories and eventual requirements for the SenseCare platform. Since participatory design involving people living with dementia could bring about unique challenges, we adopted a dyad approach where a caregiver and the person living with dementia participate together in the focus group. This ensures that their needs are fully represented and that consent is fully transparent. In this paper, we report the personal stories elicited during these discussions which will ultimately inform the implementation of the SenseCare platform.

With the large volume of text available online it is becoming impractical to use supervised machine learning methods that require a sizeable training set of labelled data. In this paper we introduced a new sentiment-topic model called the hybrid sentiment-topic model (HST). The HST model is a completely unsupervised sentiment classification method that allows for the topical context of words in documents to be accounted for when classifying sentiment. The only input needed for the model is a list of positive seed words, a list of negative seed words, and the number of topics. The HST model differs from similar models as it ensures that each objective topic discovered has both a positive sentiment-topic and negative sentiment-topic associated with it; other similar models do not guarantee symmetric sentiment-topics. The HST model performs three functions, firstly, it discovers objective topics in a corpus of text; secondly, it finds a positive and negative sentiment-topic for each objective topic; and finally, it performs sentiment classification. The HST model is tested using a dataset consisting of movie reviews and a dataset of social media posts. For each dataset a variety of seed word lists and different numbers of topics are tested; the HST model is then compared against similar sentiment-topic models. In all experiments conducted, the HST model was found to outperform similar sentiment-topic models in terms of classification accuracy by a noticeable margin. Additionally, the HST model was found to converge faster than similar models and the accuracy was found to be more stable during the Gibbs sampling process.

There is growing interest in the development of ambient assisted living services to increase the quality of life of the increasing proportion of the older population. We report on the Night Optimised Care Technology for UseRs Needing Assisted Lifestyles project, which provides specialised night time support to people at early stages of dementia. This article explains the technical infrastructure, the intelligent software behind the decision-making driving the system, the software development process followed, the interfaces used to interact with the user, and the findings and lessons of our user-centred approach.

Technology has been adopted to mitigate adverse effects associated with aging. Ambient Assisted Living solutions provide user assistance and support which is potentially both efficient and effective. This paper describes initial evaluation results for a Multi-Agent system prototype that provides assistance and support during the day and night through an interface that is adapted according to a person's requirements profile, time of day and current activity or event. Appropriate feedback based on user context is important. This includes historical feedback which may indicate trends, which may not be apparent, particularly where the user may be forgetful.

Multi-Agent Systems provide the software to analyse and understand the data emanating from sensor networks in support of Ambient Assisted Living. We report on the implementation of interfaces which are controlled and dynamically updated by such a multi-agent system. The system can respond to changes of context and tailor interventions and interactions based on the individual who is interacting with the interface.

Today's self-managing systems would ideally be able to adapt themselves (their internal structure or behavior), as well as to autonomously participate in larger, self-organizing systems. Analogously, the enterprises or other socio-economic systems autonomously manage themselves - they make decisions on how to adapt their structure and behavior, and how to organize with other entities in the environment. To connect internal self-adaptive with external self-organizational behavior, an enterprise is “aware” of itself and of its environment, and acts according to this awareness. This position paper proposes to address the challenges of a complex distributed self-managing system by making entities in such a system able to adapt themselves similarly to how companies manage themselves in socio-economic systems. To enable the knowledge transfer between these two fields, the paper proposes to utilize symbolic models which will be used by self-managing systems for knowledge representation and reasoning. This will make such systems in a way also self-aware and enable both self-adaptive and self-organizing capabilities. The paper discusses research directions to make this approach possible.

Pervasive computing services exploit information about the physical world both to adapt their own behavior in a context-aware way and to deliver to users enhanced means of interaction with their surrounding environment. The technology to acquire digital information about the physical world is becoming more available, making services at risk of being overwhelmed by such growing amounts of data. This calls for novel approaches to represent and automatically organize, aggregate, and prune such data before delivering them to services. In particular, individual data items should form a sort of self-organized ecology in which, by linking and combining with each other into sorts of “knowledge networks” (KNs), they are able to provide compact and easy-to-be-managed higher level knowledge about situations occurring in the environment. In this context, the contribution of this paper is twofold. First, with the help of a simple case study, we motivate the need to evolve from models of “context awareness” toward models of “situation awareness” via proper self-organized “KN” tools, and we introduce a general reference architecture for KNs. Second, we describe the design and implementation of a KN toolkit that we have developed, and we exemplify and evaluate algorithms for knowledge self-organization integrated within it. Open issues and future research directions are also discussed.

Assistive technology in smart homes for elderly people with Alzheimer's disease is needed to support 'aging in place'. In this paper, we propose a probabilistic learning approach to characterise behavioural patterns for multi-inhabitants in smart homes. Decision support is then provided to monitor and assist patients to complete activities of daily living (ADL). Reasoning is based on the learned profiles and partially observed low-level sensors information. Data are stored in the proposed snow-flake schema based on homeML (an XML based schema for representation of information within smart homes). A laboratory has been developed for studying activities of 'making drinks' for multiple users. Evaluations of our learning and decision support approach are carried out on both real and simulated data. The potential of our approach to support assistive living and home-health monitoring of Alzheimer's patients is demonstrated.

Adapting to current context of usage is of fundamental importance for pervasive computing services. As the technology for acquiring contextual information is increasingly available and as it is producing growing amounts of data, there is the need for tools to organize such data before delivering it to services. This produces a sort of "knowledge networks " representing comprehensive knowledge related to a "situation " in an expressive yet manageable way. In this paper, also with the help of a simple case study, we motivate the need for situation-awareness and for knowledge networks, introduce a reference architecture for knowledge networks, and exemplify a prototype implementation thereof. Finally, current and future research directions are discussed.

Autonomic computing provides mechanisms for the self-management of computing systems. This paper proposes context prediction as an autonomic mechanism to improve the usability of ubiquitous computing environments; in our case, for behavioral prediction in an environment that supports independent living for ageing people. A purpose of this paper is to stimulate debate on how best to improve ubiquitous computing environments for those who inhabit them; in particular for those ageing people who wish to continue to live independently. We propose a layered and extensible context architecture that provides self-managed and self-configuration capabilities. In particular, in addition to the context provider and context service layers, we propose context prediction and context fusion layers with cross-layer context-quality capabilities. We envisage a distributed peer-to-peer architecture with some form of semantic overlay network facilitating autonomic communications within and beyond the ubiquitous home environment.

This paper describes potential synergies between data mining and XML, which include the representation of discovered data mining knowledge, knowledge discovery from XML documents, XML-based data preparation and XML-based domain knowledge. Each category is viewed from a theoretical as well as a practical point of view.