Suicide is a complex phenomenon wherein, in addition to the individual impacted, its effects seep into many lives including those of their caregivers. Caregivers seek help everywhere but face unique challenges including limited access to timely resources and personal mental health struggles. Mobile health apps offer a promising solution, but addressing caregivers’ specific needs remains a concern. We present LifeLink, a persuasive mobile app to support caregivers of individuals experiencing suicidal thoughts. The app was developed in three stages. First, we reviewed 80 existing suicide prevention apps. Second, we designed a low-fidelity prototype of LifeLink using the Persuasive System Design model and refined it through a study conducted with 45 caregivers. Finally, incorporating evidence-based strategies and caregiver feedback, we developed and evaluated LifeLink in another study with 50 caregivers. Results show that LifeLink is user-friendly, engaging, elicits positive user experience and effectively empowers caregivers. LifeLink usage was associated with improved mental wellbeing, increased mental health literacy, and a more supportive environment. Our findings highlight the importance of involving caregivers in the design process. We offer recommendations for designers and researchers developing impactful persuasive technology for suicide prevention and for those working in related areas.

Suicide is a complex phenomenon because in addition to the individual who is impacted, its effects seep into many lives including caregivers of the individual experiencing suicidal thoughts and those who die by suicide. Often, caregivers seek help everywhere, but face various challenges including long wait times, difficulty accessing resources quickly, personal struggles with mental health, and time sensitivity. Mobile-health interventions are a promising solution for this, as they are easily accessible, available, geographic location independent and affordable. Many mobile apps for suicide prevention exist. However, not much is known about their design and efficacy, and if these apps address the unique mental health needs of caregivers dealing with suicidal individuals. To address these gaps, we designed LifeLink, a persuasive mobile application (app) specifically for supporting caregivers of individuals experiencing suicidal thoughts. Firstly, we reviewed 25 studies to determine risk factors that are most strongly associated with suicide and considerations for developing technological interventions for suicide prevention. Secondly, we reviewed 80 suicide prevention apps from app stores and academic literature to identify gaps and different persuasive strategies’ implementations using the Persuasive System Design (PSD) model. Thirdly, using the findings from the two systematic reviews, we designed a low-fidelity prototype of LifeLink app, implementing various evidence-based persuasive strategies. Next, using a user-centered design approach, 45 caregivers evaluated the user experience of LifeLink. Finally, we conducted a survey and semi-structured interviews to understand caregivers’ needs when supporting an individual, their assessment of perceived persuasiveness of the app and implemented strategies. Our results reveal that all the persuasive strategies were effective and significantly persuasive. LifeLink was found to be easy to use, engaging, useful, easy to navigate, elicited positive user experience, was helpful and impactful for caregivers. We also conducted a thematic analysis of participants’ qualitative feedback to uncover more insights. Findings from our evaluation demonstrate the potential of LifeLink as a valuable tool for caregivers that can increase mental health literacy and foster a supportive environment.

Mobile health (mHealth) apps are handy tools for tackling stigmatized mental health issues, including suicide. Mobile-based interventions for suicide prevention are easily accessible, increase the likelihood of honest reporting on sensitive topics and reduce stigma as compared to face-to-face or traditional interventions. Many mHealth apps for suicide prevention exist. However, the persuasive strategies employed in these apps and their efficacy remains unknown. To address this gap, we reviewed 80 suicide prevention apps available on app stores and in academic journals. We identified different persuasive strategies implemented in these apps using the Persuasive System Design (PSD) model. We also identified current trends within these apps, most and least-dominant implementations of persuasive strategies, effectiveness of apps, evaluation methods, and app content. We found that Personalization (n = 32) and Self-monitoring (n = 29) were the most-dominant strategies and Social Comparison, Social Role were the least-dominant strategies in suicide prevention apps. Based on our findings we discuss three major concerns in developing suicide prevention apps and offer recommendations for mitigating them. Our results show that persuasive strategies are a promising tool that can be used for designing suicide prevention apps. Our conclusions and recommendations will guide future work in suicide prevention app development and enhance the usability, effectiveness, and user-experience of such apps.

In this study, we conducted a comprehensive evaluation of user reviews of suicide prevention apps using machine learning (ML) and thematic analysis to assess usability and user experience. Given the critical public health issue of suicide, mobile applications have emerged as potential intervention tools. However, user perspectives are underexplored. Our analysis encompassed user reviews from 39 suicide prevention apps, totaling 110,338 reviews. We employed natural language processing for sentiment analysis and implemented five ML classifiers with Random Forest achieving the highest F1 score (87.19%). The thematic analysis revealed seven negative themes including access to therapy and privacy concerns, alongside five positive themes such as app functionality and user engagement. This research quantitatively and qualitatively evaluates user sentiments, providing insights that can guide developers and policymakers in enhancing the design of suicide prevention apps, thereby addressing mental health challenges in the digital landscape.

Grant-free (GF) access is considered in the uplink for efficiently supporting massive machine type communication (mMTC) scenarios in the Internet of things (IoT). Being a contention-based procedure, GF transmissions consisting of preamble followed by data are susceptible to collisions. This results in an increased number of failed users who retransmit and add to the load on the network. Given that the network is already dense, it is essential to reduce collisions and hence minimize the number of failed users. In this work, a system model for multi-user GF uplink transmission with repetitions is proposed, where the time-frequency resources are divided into transmission opportunity (TO) groups whose size is equal to the number of repetitions adopted by the users. Considering preamble transmissions at the link layer, the key performance indicators (KPIs) corresponding to the per-user and all-user success probabilities of the users in such a system are derived. Further, a method to determine the number of preambles based on the sum of the all-user success probability and the probability of a single user failing in the system is proposed, and closedform expressions for the same are derived. It is demonstrated that our proposed method to find the number of preambles ensures that the collisions (and hence the number of failed users) are reduced considering two scenarios based on the number of users arriving - a) fixed and b) random (following a Poisson distribution). The correctness of our analysis is verified through Monte-Carlo simulations in both the scenarios.

The third-generation partnership project (3GPP) has proposed the narrowband Internet of Things (NB-IoT) standard for serving Internet of Things (IoT) users. Improvements to the NB-IoT standard are being considered to efficiently support massive machine-type communication (mMTC) IoT applications. Subsequent releases of the NB-IoT standard envision to support mMTC using different access mechanisms. In this work, a grant-free (GF) nonorthogonal multiple access (NOMA) system compatible with the current NB-IoT standard is proposed, where the resources are divided into transmission opportunities (TOs). The TOs are further divided into multiple groups based on the repetitions adopted by the users in the system. Physical (PHY) layer procedures for preamble transmission/detection, channel estimation, data transmission/decoding, and acknowledgment (ACK) transmission/reception are described in detail for the proposed GF-NOMA system. Specifically, the preamble sequences are also used as spreading sequences for data transmission and three types of sequences—Zadoff–Chu, random Gaussian, and Golay sequences are explored. Results are presented for per-user success probability, latency, and energy consumption considering two NB-IoT compatible settings along with repetitions and retransmissions. It is demonstrated that the proposed GF-NOMA scheme meets the mMTC requirement recommended by the 3GPP with more than 99% per-user success probability and with the energy consumption reduced by at least 80% compared to NB-IoT, while being readily compatible with the current NB-IoT standard.

We describe an approach to task-oriented dialogue in which dialogue state is represented as a dataflow graph. A dialogue agent maps each user utterance to a program that extends this graph. Programs include metacomputation operators for reference and revision that reuse dataflow fragments from previous turns. Our graph-based state enables the expression and manipulation of complex user intents, and explicit metacomputation makes these intents easier for learned models to predict. We introduce a new dataset, SMCalFlow, featuring complex dialogues about events, weather, places, and people. Experiments show that dataflow graphs and metacomputation substantially improve representability and predictability in these natural dialogues. Additional experiments on the MultiWOZ dataset show that our dataflow representation enables an otherwise off-the-shelf sequence-to-sequence model to match the best existing task-specific state tracking model. The SMCalFlow dataset, code for replicating experiments, and a public leaderboard are available at https://www.microsoft.com/en-us/research/project/dataflow-based-dialogue-semantic-machines .