We aimed to explore the potential incremental cost-effectiveness of the PRODEMOS coach-supported mobile health intervention for primary prevention of dementia versus standard of care provided to people aged 55-75 years with low socio-economic status (SES) in the United Kingdom (UK), and any SES in China. 12-18-month PRODEMOS trial (ISRCTN15986016) efficacy outcomes on hypertension, obesity, hypercholesterolemia, physical inactivity and smoking were extrapolated to lifetime impact on dementia onset, myocardial infarction, stroke and death using a health-economic open-source simulation model. Simulated outcomes showed dementia cases were avoided (UK = -206 (-658 to 281), China = -140 (-456 to 205) per 100,000 persons) and disease-free time was gained for dementia, myocardial infarction and stroke (mean months per person UK = 0.4, 0.0 and 0.0; China = 0.2, 0.0 and 0.0 respectively). Assuming a maximum intervention duration of 10 years with a 10 % annual non-adherence rate, the incremental net health benefit in the UK (-0.190) and China (-0.009) indicated a potential lack cost-effectiveness. Our method was limited by strong assumptions regarding causality and sustained effectiveness, lack of some country-specific input estimates, and the lack of probabilistic analysis. The PRODEMOS coach-supported mobile health intervention for the primary prevention of dementia, aimed at people aged 55 to 75 years with low SES in the UK and those of any SES in China, may potentially lack cost-effectiveness in both countries. However, lack of data required strong assumptions regarding causality and sustained effectiveness, which limited policy recommendations.

The prevalence of gestational diabetes mellitus (GDM) in Malaysia is estimated at 9-18%. Although GDM is associated with increased and potentially modifiable risk of developing type 2 diabetes, the effectiveness of diabetes prevention interventions (DPI) post-GDM in this setting is unclear. To evaluate the feasibility of conducting a future full-scale, two-arm, parallel, randomised controlled trial (RCT) of a DPI in women with GDM set in Malaysia. Women in both arms received usual GDM care. Women in the intervention arm also received modules on diet, physical activity, and mental health via a mobile application, over six months post-partum, plus dietitian-led group sessions and motivational text messages. The primary feasibility outcomes included the proportion of women who consented, were eligible and randomised and provided outcome data. We measured biomedical and mental health outcomes for a full-scale RCT at four time points: baseline before randomisation (approximately 30 weeks' gestation), 36 weeks' gestation and 3- and 6-months postpartum. We screened 660 women with GDM, 294 (45%) consented for eligibility screening, of whom 164 (24.9%) were eligible and 60 (9%) consented and were randomised. The proportion who completed biomedical outcomes was 85% at each follow-up. There was no treatment effect on any other biomedical outcomes or secondary outcomes. The participation rate was in keeping with previous DPI trials and the attrition rate was low, suggesting it is feasible to conduct a full-scale RCT.

Although important learnings come from traditionally designed large prospective asthma cohorts, highly restrictive inclusion and exclusion criteria limit generalisability to clinical practice. Moreover, small sample sizes for important disease subtypes, narrow scope of clinical data collection and limited biomarker assessments reduce the power of some studies to detect important and diverse longitudinal disease courses. The Real-world and Genomic data-based Asthma Insights through Network Analysis (REGAIN) study takes a novel approach to asthma cohort development by employing a pragmatic definition of asthma and simplified study procedures for biospecimen and data collection. REGAIN will produce a large scale, real-world, longitudinal clinical and molecular description of asthma powered to characterise and compare clinically relevant asthma subtypes. This design will provide insights on distinct longitudinal trajectories of disease, predictors of response to therapies and likelihood of clinical remission, all of which should help guide asthma management. REGAIN is a clinical observational retrospective and prospective cohort study designed to determine large scale, real-world longitudinal clinical and molecular descriptions of asthma according to types of treatment, level of asthma control and inflammatory biology based on clinical biomarkers. Key questions include predictors of change in asthma control as well as timing and durability of clinical remission on biological therapy. To complement these clinical insights, REGAIN will produce one of the largest multiscale data sets in asthma that will include demographic and clinical features, inflammatory biomarkers, responses to therapy with inhaled steroids and other inhaled controllers with or without asthma biologics, and serial airway epithelium and peripheral blood transcriptomics and proteomics. REGAIN targets enrolment of 780 participants with asthma fitting one of five prespecified asthma subtypes with the aim of better characterising under-studied groups and allowing comparative analyses to elucidate important differential therapeutic responses and clinical trajectories. We target enrolment of 400 healthy controls to provide a healthy state molecular description of the tissues sampled in REGAIN participants with asthma. Participants with asthma are followed prospectively for 18 months with assessment of longitudinal clinical status including prospective clinical data collection, integration of electronic medical record data and serial biospecimen collection at 6 and 18 months. Participants with asthma starting treatment with asthma biologics undergo additional clinical assessment and biospecimen sampling at 3 months to track early clinical and molecular response to therapy. Healthy participants without asthma are evaluated cross-sectionally on enrolment without longitudinal follow-up in order to compare molecular profiles for airway epithelium and blood. An optional study component for participants with asthma employs a mobile phone application, digital inhaler monitors and home digital peak flow measurements and contributes data on real-time medication use, serial lung function and geolocated environmental data relevant to asthma. The REGAIN protocol and all amendments were approved by The Icahn School of Medicine at Mount Sinai Program for Protection of Human Subjects (PPHS19-0358), and all participants provided written informed consent. Enrolment began in November 2019 and was completed in February 2024. Results will be presented at local, national and international meetings, and results will be submitted to peer-reviewed journals for consideration for publication. NCT06623435.

Cochlear implants (CIs) enable people with moderate-to-profound sensorineural hearing loss to perceive sounds, but perceptual outcomes with a CI remain highly variable from person to person. Individuals who wear their CI consistently tend to exhibit better auditory outcomes than those who do not wear their CI as often. However, even the best-performing CI users typically exert more effort to understand speech than their peers with normal hearing, which can lead to listening-related fatigue. In this study, we aimed to understand whether the amount of time an individual wears a CI each day influences his or her listening-related fatigue and subjective hearing abilities. We hypothesized that individuals who wear their CI consistently will experience less listening-related fatigue and better subjective hearing abilities than those who do not wear their CI as often. We surveyed 71 CI listeners, including 40 adults (age 18 to 79 years) and 31 children (age 7 to 16 years). Listening-related fatigue was estimated via the adult, pediatric, and parent versions of the Vanderbilt Fatigue Scale. Subjective hearing abilities were estimated using the adult, pediatric, and parent versions of the Speech, Spatial, and Qualities of Hearing Scale. CI daily wear time was estimated using self-report measures and data from the participants' mobile phone applications. Adults who wore their devices less often during the day experienced significantly more listening-related fatigue and more physical exhaustion than those who wore their devices more consistently. Higher levels of listening-related fatigue were associated with poorer subjective hearing abilities in the adults with CIs. On the other hand, children with CIs and their parents did not report high levels of listening-related fatigue overall (i.e., their responses fell within the normative range of scores on the child and parent versions of the Vanderbilt Fatigue Scale, respectively). Daily device wear time did not predict either parent-reported or child-reported listening-related fatigue. However, higher mental fatigue was associated with poorer subjective hearing abilities in children with CIs, as reported by their parents. There were no significant relationships between child-reported listening-related fatigue and subjective hearing abilities. Adult CI users who consistently wear their device experience less listening-related fatigue and physical exhaustion than those who wear their device less often. This information can guide clinicians in counseling their adult patients to wear their devices consistently, potentially leading to improved quality of life. We did not observe relationships between device usage and listening-related fatigue in the children with CIs, likely because our pediatric participants reported relatively low levels of listening-related fatigue overall. There is a need to explore factors that contribute to listening-related fatigue in children with CIs who experience a broad range of hearing histories and fatigue symptoms.

Abstract Introduction Gradual dosage reduction (tapering) is the recommended approach for reducing and/or stopping the use of psychiatric medication in order to minimise the risk of withdrawal symptoms. A previous James Lind Alliance Priority Setting Partnership highlighted several gaps in the current evidence base on tapering psychiatric medication.[1] To date, the evidence base for mobile phone applications (‘apps’) and app-based interventions in supporting safe and effective discontinuation of psychiatric medications has not been examined. Aim To examine the content, underpinning evidence base, and impact of mobile phone apps and app-based interventions that focused on supporting the tapering of psychiatric medication. Methods A scoping review was conducted using the Joanna Briggs Institute guidance and results were reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews guideline.[2] Seven electronic databases (MEDLINE, EMBASE, CINAHL, PsycINFO, Web of Science, ACM, IEEE Xplore) and two app stores (Apple’s App Store (iOS) and Google Play Store (Android)) were searched. To meet inclusion criteria, studies of any design identified from peer-reviewed literature (i.e. ‘non-commercial’ apps) and apps identified from searches of commercial app stores (i.e. ‘commercial’ apps) had to focus on supporting the tapering of psychiatric medication. Following screening, key information was extracted from included studies and apps. Identified apps were coded in terms of their functionalities and components of the Behaviour Change Technique Taxonomy version 1 (BCTTv1). The findings were described using narrative synthesis. Results Seven apps met inclusion criteria comprising two non-commercial apps, and five commercial apps. Two non-commercial apps were evaluated using randomised controlled trials (RCTs); one RCT was ongoing and the other was stopped prematurely due to recruitment challenges. All seven apps focused on supporting the tapering of psychiatric medication. Eight functionalities and seventeen BCTs were identified across the seven included apps and app companion websites. The four most common BCT codes were: 1.3 ‘Goal setting (outcome), 2.3 ‘Self-monitoring of behaviour,’ 4.1 ‘Instructions on how to perform the behaviour,’ and 8.7 ‘Graded tasks.’ Most of the identified apps did not provide information about the app development process or specify any theoretical underpinning. Given the lack of reported outcomes for the evaluation studies of app-based interventions and commercial apps, it was not possible to report the impact from a qualitative or quantitative perspective. Conclusion This scoping review demonstrated a paucity of mobile phone apps and studies of app-based interventions that support tapering psychiatric medication. The review’s main strength was that it followed a published protocol, and data extraction and app coding was conducted by two reviewers independently. In terms of limitations, the lack of studies evaluating app-based interventions made it difficult to address the objective relating to the impact of the identified apps. By outlining the key functionalities offered by the identified apps and characterising them in terms of component BCTs, as well as raising awareness of the shortcomings of commercially available apps, the review findings will guide future research on the development and evaluation of apps to support individuals who want to reduce or stop psychiatric medication use.

This study visits the effects of a new smartphone application on outcome reporting and disease activity management in patients with ankylosing spondylitis, a long-term inflamed disease intervention. The main goal was to determine whether digital self-monitoring can help improve patient-reported outcomes and gains better medical adherence. A 12 week prospective observational study was conducted with 77 patients who will use the app to report weekly BASDAI scores and medication logs. Statistical analysis of paired t-tests, ANOVA, Pearson correlative and Cronbach’s alpha were conducted through SPSS v27. As shown from the results, there was a notable decline in the mean BASDAI scores was observed with greater improvements among the patients more adherent. Positive correlation (r = 0.62, p < 0.01) was found between adherence rates and symptom reduction. The internal consistency of app based PRs was high throughout the study(mean α = 0.88). These discoveries imply that mobile phone applications are able to function as dependable digital options for pharmacological surveillance and individualized disease control in ankylosing spondylitis.

Abstract Background Hand trauma accounts for nearly one third of Emergency Department (ED) attendances in the United Kingdom (UK). Flexor tendon repair is one of the most commonly performed Plastic Surgery procedures requiring months of repeat contact with hand therapy for optimal outcomes, the environmental impacts of which have not previously been evaluated. Methods A retrospective study of 17 patients undergoing flexor tendon repair. Process mapping and life-cycle analyses of the patient pathway, from ED to surgery, to rehabilitation and full recovery, was performed. Estimates for carbon dioxide emissions were generated in kgCO2eq. allowing identification for specific opportunities for pathway optimization, in line with national guidelines. Race and ethnicity were not reported in this study. Results This study estimated the carbon footprint of a patient undergoing flexor tendon repair to be approximately 113.2 kgCO2eq. The rehabilitation phase made the greatest contribution to emissions (59.85 kgCO2eq., 52.9%) and patient travel was responsible for a staggering 79.42% of total emissions (89.88 kgCO2eq.), followed by the production and transport of equipment (8.52 kgCO2eq., 7.53%) needed for treatment. Conclusions To reduce carbon emissions, reducing the need for patient travel was suggested. This can be achieved by greening hand therapy through increased use of virtual consultations and encouraging the use of mobile phone applications. Adoption of circular economy principles through the use of reusable surgical gowns and drapes, creation of procedure-specific equipment packs, and increased use of wide-awake local anaesthesia, no tourniquet (WALANT) was also recommended.

Touchscreens and physical remotes work fine—until they don't. In a hospital prep room, a wheelchair user's kitchen, or anywhere shared surfaces carry infection risk, they become an actual problem rather than a minor inconvenience. This work describes a gesture-controlled IoT system we put together using an ESP32-CAM as the capture device, a MediaPipe-plus-CNN-LSTM pipeline for recognition, and MQTT for getting commands to devices. The camera is cheap, wireless, and small enough to tuck into a corner without anyone noticing it. Recognised gestures reach lights, HVAC controls, or media systems in roughly 174 ms. A React.js dashboard on a local server keeps track of everything—what the camera currently sees, what commands went out, device states, usage patterns over time. There is also a native Android app built in Android Studio that shows the same information on a phone and sends Firebase push notifications when something worth knowing about happens. We ran the whole system continuously for thirty days in a test room fitted with real smart devices. Accuracy across ten gesture types came out at 94.7%. The two gesture classes that kept underperforming during the first week had, by day 30, closed most of the gap on their own—the background retraining loop gathered new samples for those classes only and quietly updated the weights, never touching the gestures that were already working well.

This study explores the added value of co-design in addition to other innovation research methods in the process of developing a serious game design document for a road safety game. The sessions aimed at exploring 4 aspects of a location-based game experience: themes, game mechanics, mobile phone applications and locations for mini-games. In total, 72 adolescents between 15 and 18 years participated in five co-design sessions lead by a researcher and a professional game designer. The sessions provided useful input regarding the aspects the authors wished to explore. The sessions were especially useful in gathering input on scoring systems, ways to give instructions about next tasks and organizing level systems. In sum, their study indicates that co-design can be a source of additional ideas on top of other research methods such as state of the art analysis and expert consultation and thus lead to more effective interactive content creation.

This paper presents a mechanical simulation of image sensor packages, with a particular emphasis on accurately modeling the cure shrinkage and viscoelasticity. Image sensors have a wide application in mobile phones, autonomous vehicles, and medical imaging. They are highly sensitive to warpage and misalignment, which can significantly degrade the image quality. The encapsulant acts as a primary barrier, protecting the sensor from external factors such as moisture and thermal mechanical stress. Accurately modeling and predicting the package warpage is crucial to ensure the optimal optical performance. Previous simulation studies on image sensor packaging often neglect cure shrinkage, leading to inaccurate warpage predictions. Cure shrinkage, a phenomenon where the material contracts during the curing process, can induce significant stress on the package and lead to warpage in addition to CTE-mismatch. This work developed a new simulation approach to incorporating cure shrinkage modeling, and it achieved a much higher degree of accuracy in predicting package warpage, as demonstrated by the close correlation between simulation results and actual warpage testing data. Furthermore, this paper studied the impact of using viscoelastic properties and compared time-dependent deformation with elastic solutions. Finally, this paper conducted comprehensive design of experiments (DOE) studies to evaluate the impact of different encapsulant materials on the susceptibility to glass cracking. The findings of this work are particularly useful for improved warpage prediction in simulation and better understanding of encapsulant properties, ultimately optimizing the package design and enhancing the reliability and longevity of electronic devices.

Hand trauma accounts for nearly one-third of emergency department (ED) attendances in the United Kingdom (UK). Particularly, repair of flexor tendon injuries is one of the most commonly performed plastic surgery procedures and requires months of repeated contact with hand therapy for optimal outcomes. The environmental impacts of treatment have not previously been evaluated. A pilot retrospective service evaluation using process mapping and life-cycle assessment of 17 patients undergoing flexor tendon repair was performed using the functional unit of one completed patient pathway cycle, from ED to surgery, to rehabilitation and full recovery. Data were collected retrospectively from electronic patient records, supplier data and emission factor databases. Estimates for carbon dioxide emissions were generated in kgCO2eq., allowing us to look for specific opportunities for pathway optimisation, in line with the national guidelines. We estimated the carbon footprint of a patient undergoing flexor tendon repair to be approximately 113.2 kgCO2eq. The rehabilitation phase made the greatest contribution to emissions (59.9 kgCO2eq., 52.9%) and patient travel was responsible for 79.4% of total emissions (89.9 kgCO2eq.), followed by the production and transport of equipment (8.5 kgCO2eq., 7.5%) needed for treatment. The results of this pilot eco-audit identified key emission sources and potential interventions to reduce carbon emissions for this pathway. These include reducing the need for patient travel, which can be achieved by greening hand therapy through increased use of virtual consultations and encouraging the use of mobile phone applications. We also recommend the adoption of circular economy principles by using reusable surgical gowns and drapes, creation of procedure-specific equipment packs, and increased use of wide-awake local anaesthesia, no tourniquet.

BACKGROUND:Self-care programs based on mobile phone applications facilitate the management and control of chronic obstructive pulmonary disease (COPD), and identifying self-care needs is necessary for designing such programs. This study aims to identify the needs of COPD patients to design a self-care program based on a mobile phone application.MATERIALS AND METHODS:In this multi-method study, the self-care needs of COPD patients were identified with two approaches: literature review and quantitative content analysis. Articles were extracted through a comprehensive review of studies from the past 10 years. Self-care needs were identified via semi-structured interviews with 18 participants.RESULTS:In the literature review, out of 1,671 articles retrieved, 27 were selected by removing duplicate and unrelated articles. In the qualitative research, three categories of educational needs were identified: the maintenance of psycho-emotional health, physical health, and effective communication and support. After combining the extracted needs, the patients’ self-care requirements were categorized into three areas: physical health, enhancement and promotion of psycho-emotional well-being, and effective communication and connection with healthcare professionals, family, friends, and community.CONCLUSION:The needs identified in the present study were divided into three groups: the need to improve and enhance psycho-emotional status; the need to improve physical status; and the need to establish effective communication with healthcare professionals, family members, and community; which can be used in designing a self-care program based on a mobile application.

In this study, an annular laser beam shaping optics and a wire feeding system are used for additive manufacturing. A discrete concentric layering trajectory strategy (DCL-TS) and a continuous deposition trajectory strategy (CD-TS) for the laser-directed energy deposition (WL-DED) of aluminum alloy stud structures are developed. Initially, combinations of parameters, such as laser power, transverse speed, and wire feeding speed, which lead to a process that produces a single-layer structure with good morphology and no visible pores and cracks, are identified. Then, DCL-TS and CD-TS manufacturing strategies are used to produce aluminum alloy studs of similar dimensions. The EBSD results indicate that the CD-TS produces finer grains in the aluminum alloy studs compared to the DCL-TS; correspondingly, mechanical testing reveals superior microhardness and tensile strength in the circularly fabricated studs. The latter tensile value testing verifies that aluminum alloy studs using WL-DED on the substrate can meet the requirements for practical application in mobile phones, computers, etc. This research method enhances the mechanical properties of additively manufactured items. Consequently, manufacturing efficiency is significantly improved, providing a promising solution for rapid production.

With the modernization of the Internet and society, people often use some mobile phone applications to facilitate their lives. But since most of these apps require sensitive information from users, malware has been developed by malicious people to infect users' phones, especially on the Android platform. In order to detect whether a file after download is malware, this paper takes a static detection method. By decompiling and uncompressing the downloaded APK file, extracting the software rights from it, converting the rights into feature graphs, and putting them in a neural network to test. After four fold cross-testing, the average accuracy on the validation set was 88.205 percent. The group that trained the network with the highest accuracy of its data sets was then selected as the training set for the final network, and then tested on a test set, which showed that this method worked better than traditional machine learning methods.

The use of plant protection products (PPPs) in farming systems poses a potential risk to honey bee health, yet beekeepers’ awareness and interactions with farmers remain understudied. This study compares the knowledge, perceptions, and PPP risk mitigation practices of amateur and professional beekeepers in Ireland and Greece, based on survey responses from 476 participants. Irish beekeepers were predominantly amateurs, while most Greek beekeepers were professionals. Irish amateurs reported significantly less experience in beekeeping and managed fewer hives compared to the other groups. Across both countries, insecticides were perceived as the most harmful PPPs to honey bees, with Greek beekeepers expressing greater concern than Irish counterparts. Greek beekeepers showed greater confidence in identifying PPP poisoning and were more likely to conduct PPP residue testing. Most beekeepers had limited interaction with farmers and were rarely notified before PPP application, often too late to take protective measures. Risk mitigation practices were minimal among Irish beekeepers, while Greek professionals more actively avoided agricultural areas. Beekeepers strongly agreed that farmers should be better informed on the risks of PPPs and expressed a desire for improved communication, preferably via mobile phone applications. Most respondents indicated interest in receiving further education, ideally through beekeeping associations. The findings highlight the need for targeted educational programs, structured beekeeper-farmer communication and collaboration, and accessible tools to mitigate PPP-related risks to honey bees. These insights can support the development of more effective pollinator protection and agro-ecological cooperation strategies.

Background: Diabetes mellitus is a metabolic disease that is a global problem, with the number of sufferers increasing annually. Adherence to the five pillars of diabetes management is the cornerstone of preventing and reducing morbidity and mortality from diabetes. Purpose: To analyze how to improve adherence in type 2 diabetes patients in implementing the five pillars of diabetes management. Method: The research method used a literature review approach to map and identify interventions for implementing the 5 pillars of diabetes management. Results: The study found 20 articles describing interventions for people with diabetes mellitus through educational and informational delivery mechanisms. The types of interventions included: mobile phone application interventions, physical exercise interventions, audiovisual interventions, flipchart interventions, and peer leader interventions. Conclusion: Five types of interventions improve adherence in patients with type II diabetes mellitus to implementing the 5 pillars of diabetes management.

As adult educators in Vancouver's Downtown Eastside, public health workers and those with lived experience of drug use create supportive environments and educational programs that foster harm reduction and community empowerment to address the Opioid Crisis. Using Community Empowerment theory as its conceptual framework and an Appreciative Inquiry with public health experts as its methodology, this research study contributes to understanding community-led adult education initiatives for marginalised adults, specifically in public health education. Through this exploration, we examine the working relationships and aspects of community culture, public education, social media and cell phone apps employed for education, communication, and life-saving tools to promote positive social change. Findings provide an overview of the educational work currently being done by public health adult educators, the community, and activist users of drugs and highlight the need to address social determinants of healthcare, including systemic oppression and access to internet technology for community empowerment.

The proliferation of mobile commerce channels has fundamentally reshaped retail ecosystems, particularly in digital markets like China where smartphone adoption approaches saturation among internet users. While extant literature has extensively examined the impact of new channel introductions (e.g., online, offline, mobile) on firm performance, less attention has been paid to consumer behavioral nuances within established digital interfaces. Addressing this gap, our study pioneers a comparative analysis of purchasing dynamics across two dominant yet technologically distinct channels: native apps versus light-app channels (e.g., WeChat mini-programs). While both channels share core mobile attributes (e.g., small screen sizes, on-the-go accessibility), their divergent technological architectures (Swift, Kotlin, and Java vs. HTML, CSS, WXML, and WXSS) create systematically differentiated consumer experiences. Through econometric analysis of 185,437 transaction records from a multichannel B2C platform, we reveal that consumers tend to spend more, purchase more items, and exhibit a lower likelihood of product returns when shopping through the light-app channel compared to native apps. More importantly, these behavioral divergences are moderated by product categories, price levels, and discount depths. Our findings contribute to the multichannel retailing literature by providing new insights into consumers’ behavioral differences between the two popular, yet distinct, mobile channels. Based on these insights, we suggest that multichannel retailers should prioritize channel convenience and accessibility and reconsider their investments in mobile native apps. Additionally, retailers should tailor assortments, pricing, and discount strategies to each channel to effectively engage consumers and stimulate purchases. Our research also emphasizes the importance of aligning marketing, operations, and finance strategies in multichannel retailing.

Facial features hold information about a person’s emotions, motor function, or genetic defects. Since most current mobile devices are capable of real-time face detection using cameras and depth sensors, real-time facial analysis can be utilized in several mobile use cases. Understanding the real-time emotion recognition capabilities of device sensors and frameworks is vital for developing new, valid applications. Therefore, we evaluated on-device emotion recognition using Apple’s ARKit on an iPhone 14 Pro. A native app elicited 36 blend shape-specific movements and 7 discrete emotions from healthy adults. Per frame, standardized ARKit blend shapes were classified using a prototype-based cosine similarity metric; performance was summarized as accuracy and area under the receiver operating characteristic curves. Cosine similarity achieved an overall accuracy of , exceeding the mean of three human raters (; percentage points, ≈16% relative). Per-emotion accuracy was highest for joy, fear, sadness, and surprise, and competitive for anger, disgust, and contempt. AUCs were ≥0.84 for all classes. The method runs in real time on-device using only vector operations, preserving privacy and minimizing compute. These results indicate that a simple, interpretable cosine-similarity classifier over ARKit blend shapes delivers human-comparable, real-time facial emotion recognition on commodity hardware, supporting privacy-preserving mobile applications.

Artificial Intelligence-Based Algorithms Improve Care of Patients with AAA.