JMIR Rehabilitation and Assistive Technologies

AI Scribe in Specialized Rehabilitation: Usability Study of Perceived Usability and Usefulness Among Physicians and Psychologists

2026-11-06T18:00:03-05:00

Background: Clinical documentation in specialized, interdisciplinary rehabilitation is often time-consuming and complex and may contribute to clinician burnout and reduced time for patient interaction. AI-based scribes transcribe and summarize clinical notes and may reduce documentation burden; however, their usefulness and usability in specialized rehabilitation remain uninvestigated. Objective: This study aimed to assess the use of an ambient AI scribe tool among physicians and psychologists in specialized rehabilitation, focusing on perceived usefulness, usability, note quality, impact on patient interaction, perceptions of time spent on documentation, and documentation-related fatigue. Methods: Over approximately 4 weeks, physicians and psychologists at 9 specialized rehabilitation centers used an ambient AI scribe during clinical consultations. Usability and user experience were assessed using the System Usability Scale (SUS) and a modified technology acceptance model (TAM). Participants rated an AI-generated note using a modified 8-item Physician Documentation Quality Instrument (PDQI-8) questionnaire to evaluate documentation quality. Additional questions explored perceived impact on patient interaction, documentation-related fatigue, and general attitudes toward AI technology. Results: Thirty-one clinicians (18 physicians and 13 psychologists) used the AI scribe over approximately one month and completed evaluation questionnaires. The mean SUS score was 78.3 (SD 10.4), with 86.7% of participants scoring above the usability threshold. Perceived usefulness (PU) and perceived ease of use (PEOU) were relatively high (mean PU score 5.65, SD 1.33; mean PEOU score 5.56, SD 1.17), and 87.1% expressed intention to use the AI scribe in future practice. Most participants (80.6%) agreed that using the AI scribe reduced time spent on documentation, while 66.6% reported a reduced perception of documentation-related fatigue. The quality of AI-generated notes was rated relatively high, with a mean modified PDQI-8 score of 4.3 (SD 0.3) out of 5. Comprehensibility received the highest rating (mean 4.6, SD 0.6), whereas thoroughness received the lowest rating (mean 3.9, SD 1.0) among the note quality variables. Overall, 71.0% of participants expressed some level of agreement that the AI scribe enhanced attentiveness during patient consultations, while a similar proportion (70.9%) agreed that it made active listening easier by reducing distractions associated with note-taking. A few participants reported that some notes were incomplete, lacked biopsychosocial details, and contained inaccuracies. Conclusions: This study demonstrated broad acceptance of an AI-based ambient scribe among physicians and psychologists working in specialized rehabilitation. The AI scribe was generally perceived as valuable, easy to use, and clinically relevant. Most participants reported a reduced perceived documentation burden and improved attentiveness during consultations. These findings suggest that clinicians perceived the AI scribe as supporting more focused patient interactions. Some concerns were noted regarding completeness and occasional inaccuracies, highlighting areas for refinement. Future research should explore long-term implementation in routine rehabilitation settings and its effects on workflow, documentation quality, and patient experience.

Development, Usability, and Feasibility of a Pain-Stratified Smartphone-Based Home Intervention System for Patients With Adhesive Capsulitis: Exploratory Feasibility and Usability Trial

2026-11-05T16:30:23-05:00

Background: Frozen shoulder causes prolonged pain and restricted range of motion that impair daily function. Home exercise is an indispensable part of self-management, but adherence is often suboptimal under traditional printed or verbal instruction because of the lack of objective monitoring and individualized progression. Existing rehabilitation apps offer sensor-based assessment and feedback but generally lack an adaptive mechanism that tailors the exercise prescription to a patient’s dynamically changing clinical status. Objective: This study aimed to develop a personalized, pain-stratified telerehabilitation app for patients with frozen shoulder; to evaluate its usability, feasibility, and safety as an add-on to usual care; and to explore the system’s clinical outcomes on a descriptive basis. Methods: An Android app integrating smartphone sensor-based range of motion self-assessment, pain-based exercise stratification, usage records, and exercise reminders was developed. In an exploratory, assessor-blinded, randomized controlled feasibility trial, 10 patients with frozen shoulder were randomly allocated to an app group (n=5) or a handout group (n=5), and both underwent a 4-week home exercise intervention with identical content. Primary outcomes included system usability (System Usability Scale) and feasibility (retention and exercise adherence). Secondary outcomes focused on the exploratory clinical signs included pain and function (Shoulder Pain and Disability Index) and range of motion (measured by goniometer). Given the small sample size, continuous variables were analyzed nonparametrically and interpreted strictly as descriptive clinical trends without inferential hypothesis testing. Results: All 10 participants completed the trial (attrition rate 0%). The mean System Usability Scale score in the app group was 87.5 (SD 5.86), corresponding to an “excellent” usability rating. Over the 28-day intervention, the app’s usage logs objectively documented a mean of 18.0 (SD 5.0) of 28 exercise days (18/28, 64.3%) in the app group, demonstrating the feasibility of objective home monitoring. Both groups improved in Shoulder Pain and Disability Index scores and range of motion. Notably, all participants in the app group achieved a descriptive within-group improvement in the clinically challenging shoulder external rotation, whereas the handout group showed smaller marginal gains; however, between-group comparisons across all parameters were interpreted descriptively without significance testing due to the sample size. On the other clinical outcomes, the handout group showed numerically comparable or larger improvements. Conclusions: This study shows that a pain-stratified, personalized telerehabilitation app is highly usable and feasible as an add-on mobile health intervention for patients with frozen shoulder and that it provides objective home monitoring of exercise adherence that traditional handouts cannot. These exploratory findings highlight the need for a fully powered confirmatory trial. To robustly evaluate clinical efficacy, future study designs must include a common objective adherence measure, stratification variables, and an extended follow-up period.

Analysis of Health Care Providers' Attitudes and Expectations on Intradialytic Exergaming: Qualitative SWOT Study

2026-11-04T17:45:20-05:00

Background: Physical inactivity significantly contributes to the rapid decline in clinical and health-related outcomes among patients undergoing hemodialysis. Exergaming may offer a safe, effective, and sustainable solution for maintaining adequate physical activity levels in these patients. Objective: This study aims to support its clinical applicability; our aim was to examine the attitudes and expectations of health care providers regarding intradialytic exergaming (IEG) in hemodialysis care in Slovakia. Methods: We conducted 21 individual, in-person, semistructured interviews with health care providers. Using purposive sampling, we included training assistants, nurses, nephrologists, and managers from 2 dialysis centers in Slovakia. The verbatim transcripts of interviews were qualitatively analyzed using an inductive thematic analysis approach. The identified codes and themes were further deductively categorized into the strengths, weaknesses, opportunities, and/or threats quadrants of the Strengths, Weaknesses, Opportunities, and Threats (SWOT) framework. Results: We identified 2 strengths, 2 weaknesses, 3 opportunities, and 3 threats of IEG. Strengths included health care providers’ enthusiasm and belief in the safety of IEG. Weaknesses involved concerns about the complexity of IEG and low digital literacy among patients and providers. Opportunities were seen in improving information availability and specificity, adjusting clinical culture, and tailoring exercises. Threats included skepticism about the availability of financial resources for IEG, resistance to changes in clinical practice, and issues related to patient behavior and privacy. Conclusions: Health care providers perceived IEG as a safe and appropriate intervention for clinical practice. To improve implementation strategies, it is important to consider specific expectations regarding resource availability, as well as the technological and organizational aspects of IEG. Trial Registration: ClinicalTrials.gov NCT03511924; https://clinicaltrials.gov/study/NCT03511924 International Registered Report Identifier (IRRID): RR2-10.1097/MD.0000000000015570

Advances in Machine Learning Paradigms for Reliable Swimming Pool Drowning Detection: Systematic Review of Methods and Implementations

2026-11-03T16:30:18-05:00

Background: Drowning is a leading cause of unintentional injury-related mortality worldwide, particularly in swimming pool environments, where rapid detection is critical to prevent fatalities. Traditional monitoring approaches rely on lifeguards and rule-based systems, which are often limited by human error, delayed response, and environmental challenges such as occlusion, glare, and multiple swimmers. Machine learning (ML) and computer vision methods have emerged as promising alternatives for automated drowning detection; however, their methodological rigor, real-world applicability, and reliability remain unclear. Objective: This study aimed to systematically review advancements in ML-based drowning detection systems over the past 15 years. The review focused on detection methods, sensing modalities, evaluation metrics, and validation practices. It also examined how studies addressed real-world challenges such as occlusion, lighting variability, and multiswimmer environments, and identified key gaps related to generalizability, reliability, and privacy. Methods: A systematic literature review was conducted using major electronic databases, including IEEE Xplore, Scopus, PubMed, Web of Science, ACM Digital Library, and Google Scholar. A total of 4927 records were initially identified and screened by 2 independent reviewers using predefined inclusion and exclusion criteria. A final set of 30 studies was included for data extraction and synthesis. The methodological quality and risk of bias were assessed using the Prediction Model Risk of Bias Assessment Tool (PROBAST), while reporting transparency was evaluated using the TRIPOD (Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis) guidelines. Data analysis was conducted in R. Results: The findings indicate that drowning detection research is dominated by supervised, vision-based approaches (23/30, 76.7%), with convolutional neural networks (5/30, 16.7%) and You Only Look Once–based architectures (7/30, 23.3%) being the most commonly used methods. While reported performance is generally high, with mean accuracy exceeding 90.9%, evaluation practices are heavily centered on accuracy, with limited reporting of precision, recall, false alarm rate, and latency. One of 30 (3.3%) studies reported validation under real-world operational conditions, with an additional 5 (16.7%) studies using real-world data without formal real-world validation, and no studies conducted cross-dataset evaluation. In addition, critical real-world challenges such as occlusion, glare, and crowding were rarely addressed (<10% of studies), and all studies relied on private datasets. The PROBAST assessment further indicated a high risk of bias in 96.7% (29/30) of studies, primarily due to limited validation and nonrepresentative data sources. Conclusions: ML-based drowning detection systems have demonstrated substantial progress in algorithm development, particularly in vision-based models. However, this progress is not matched by equivalent advances in validation, robustness, and real-world applicability. The field is characterized by a significant gap between experimental performance and deployment readiness. Future research should prioritize standardized evaluation frameworks, external validation, comprehensive performance reporting, and the integration of multimodal and privacy-preserving approaches to enable reliable real-world implementation.

Validation of Single-Camera MediaPipe BlazePose for Knee Joint Angle Measurement Across Multiple Functional Tasks: Concurrent Validity, Interrater Reliability, and Exploratory Test–Retest Reliability Study

2026-11-03T15:30:24-05:00

Background: Single-camera markerless pose estimation may support accessible knee movement assessment, but performance can depend on task, joint angle, and detection quality, and evidence across complementary measurement properties remains limited. Objective: This study aimed to evaluate agreement of MediaPipe BlazePose Heavy knee-flexion estimates with goniometer-set static target positions and with Kinovea 2D video annotation for dynamic tasks, to assess interrater agreement of the Kinovea annotations, and to explore repeated-session performance in a small retest subsample. Methods: Squat, sit-to-stand, and bilateral single-leg stance tasks were performed by 15 healthy adults during standardized lateral-view smartphone recording. Static knee flexion was assessed at 4 goniometer-set targets (30°, 60°, 90°, and 120°). Native hip-knee-ankle included angles were converted to clinical knee flexion (0° at full extension). Agreement was assessed using the intraclass correlation coefficient (ICC), specifically ICC, signed bias and 95% limits of agreement, mean absolute error (MAE), and root mean square error (RMSE). Primary intervals used participant-cluster bootstrapping, and limits of agreement accounted for repeated measurements. Dynamic primary estimates were conditional on BlazePose target-frame confidence ≥0.3; low-confidence observations were also reported as detection failures and included in an all-output stress test. Four participants returned after 21 days to 29 days (median 25 d) for exploratory repeated-session analyses. Results: Dynamic measurement was available in 177 of 210 observations (84.3%); 33 of 210 observations (15.7%) failed the confidence criterion. Conditional task-specific ICCs were 0.773 (95% CI 0.564 to 0.875; n=71, 15 participants) for squat, 0.632 (95% CI 0.455 to 0.726; n=58, 12 participants) for sit-to-stand, and 0.136 (95% CI −0.078 to 0.482; n=48, 15 participants) for combined single-leg stance. Task-specific availability was present for 71 of 75 observations (95%), 58 of 75 observations (77%), and 48 of 60 observations (80%), respectively. Without the quality gate, squat ICC was 0.428, and RMSE was 25.43°. Static agreement with set targets resulted in an ICC of 0.874 (95% CI 0.770 to 0.953), with RMSE of 15.49° and angle-dependent signed error. Kinovea interrater ICC was 0.999 (95% CI 0.998 to 0.999) overall, which did not establish 3D accuracy. Repeated-session estimates with the 4 participants remained exploratory. Conclusions: With dynamic measurement available in 177 of 210 observations (84.3%), agreement among retained outputs depended on task, flexion angle, and detection quality. The findings support further feasibility and screening-level evaluation under controlled acquisition with per-recording quality checks; they do not support replacement of clinical goniometry, precise longitudinal decision-making, absolute 3D kinematic accuracy, or generalization to clinical populations or other smartphones.

Digitally Assisted Training System for Home-Based Scoliosis Therapy: Proof-of-Concept and Technical Verification Study

2026-10-29T11:30:22-04:00

Background: Schroth therapy is one of the conservative methods used to treat scoliosis. It includes not only therapeutic exercises in physiotherapy facilities but also concepts for independent training at home. Consistent and correct performance of this training has a major impact on the success of the therapy. However, without guidance and supervision by a therapist, patient motivation may decline. In addition, incorrect training techniques may be used, which can have a negative impact on the success of the therapy. Objective: This study aimed to develop a digitally assisted training system for home-based scoliosis therapy and to verify it technically on a laboratory scale. Methods: On the basis of training devices used in Schroth therapy, such as mirrors, balance boards, wall bars, and exercise sticks, we developed a concept for a modular, digitally assisted scoliosis training system (iScoolio). The development process included a requirements specification, iterative design drafts, and the transfer of the concepts into a functional demonstrator. Finally, a theoretical evaluation of the use cases and a technical verification of the measurement functions were carried out using specifically developed test scenarios. Results: A demonstrator with tracking, training, and modification functions was developed, consisting of 4 modules. Technical verification of the integrated measurement functions showed that the angle measurement in motion tracking (n=50) had a mean deviation of 2.29° (SD 1.66°) compared to the ARAMIS 3D optical reference system. Respiration tracking enabled the detection of all simulated respiratory cycles (n=40) and the significant distinction between shallow and deep breathing (=−27.4; <.001). The investigation of the 2 force sensor concepts in the handles (²=80.0; <.001) and supporting feet (²=50.0; <.001) showed a significant differentiation between 3 force levels, but approximately linear drift behavior in each case. When tracking the stance width (n=50), the mean value was 15.4 (SD 0.54) cm. Verification of the center of pressure measurement function (n=50) showed that 48% of the measurements had a deviation of ≤10 mm, and 82% of the measurements had a deviation of ≤20 mm. Conclusions: In this paper, we have presented the iScoolio concept, which has the potential to support home-based scoliosis therapy through adaptable devices and tracking elements. Integrated sensors enable monitoring of movements, breathing, forces, and pressure distribution during training. An associated web application visualizes the recorded data and serves as a guide for performing the exercises. Technical verification indicates that the integrated tracking methods are functional; however, there is still room for optimization in terms of hardware and software. Further validation is required before conducting clinical trials under real-world conditions.

Integrating Functional Cognitive Rehabilitation With Activity-Based Stroke Telerehabilitation: Intervention Design and Development Study

2026-10-27T13:00:29-04:00

Background: Many stroke survivors experience chronic cognitive deficits that, together with motor impairments, negatively impact well-being and daily function. Critically, poststroke cognitive deficits can interfere with successful engagement in rehabilitation programs aimed at improving participation in home and community activities. To address this issue, we developed a new telehealth intervention that integrates brief, functional cognitive rehabilitation (COG) into an existing upper extremity-focused, activity-based occupational therapy (OT) telerehabilitation program. Objective: To describe the development and specification of a telehealth intervention (COG+OT) that integrates functional cognitive rehabilitation into activity-based stroke telerehabilitation. Methods: COG+OT was developed using a multi-disciplinary, theory-informed process guided by principles of functional cognitive rehabilitation. It is reported according to the Template for Intervention Description and Replication checklist for rehabilitation (TIDieR-Rehab). The intervention was developed by a team with expertise in stroke rehabilitation from the fields of OT, neuropsychology, clinical psychology, and speech-language pathology. Cognitive rehabilitation content was adapted from an evidence-based compensatory strategy program and condensed into a brief “primer” format. The COG component was iteratively integrated into an existing upper-extremity, activity-based OT telerehabilitation program to support engagement and transfer of strategies to real-world functional tasks. In this paper, we describe the intervention structure, content, delivery, and tailoring procedures. Results: COG+OT is an 8-week, 13-session telehealth intervention delivered in three phases: (1) cognitive strategy training (“primer”; sessions 1-4), (2) review and guided application of strategies within OT activities (sessions 5‐10), and (3) integration of strategies to support generalization to everyday contexts (sessions 11-13). The intervention combines individualized, strategy-based cognitive rehabilitation with goal-oriented OT using a titrated approach. Patient-facing materials, structured session workflows, and therapist-guided metacognitive strategies support tailoring and real-world application. Conclusions: COG+OT is a theory-informed, user-centered intervention that provides a structured approach for integrating functional cognitive rehabilitation into stroke telerehabilitation. This work establishes a well-specified, replicable intervention model and supports future feasibility, effectiveness, and implementation studies. Trial Registration: ClinicalTrials.gov NCT06555302; https://clinicaltrials.gov/study/NCT06555302

Effects of Telerehabilitation on Quality of Life, Physical Capacity, and Symptom Burden Within the Post–COVID-19 Condition: Systematic Review and Meta-Analysis of Randomized Controlled Trials

2026-10-27T08:15:02-04:00

Background: Post–COVID-19 condition (PCC) is characterized by persistent symptoms and functional limitations that impair quality of life, and rehabilitation is an important component of its management. Because access to conventional face-to-face rehabilitation may be limited, telerehabilitation has emerged as an accessible alternative. However, the effectiveness of telerehabilitation across clinical outcomes remains uncertain, and the influence of different delivery modes is not well understood. Objective: This study aimed to evaluate the effects of telerehabilitation on health-related quality of life (HRQoL), physical capacity, and symptom burden in adults with PCC, and to determine whether delivery mode (synchronous, asynchronous, or hybrid) modified these effects. Methods: Seven databases (PubMed, Cochrane Library, Web of Science, Scopus, Embase, EBSCO, and PEDro) were searched for randomized controlled trials published from January 2020 through December 2025. Eligible trials enrolled adults with PCC and compared telerehabilitation with usual care or another control condition. The primary outcome was HRQoL; secondary outcomes included functional capacity assessed by the 6-minute walk test and sit-to-stand test, dyspnea, fatigue, and symptoms of anxiety and depression. Risk of bias was assessed using the Cochrane Risk of Bias 2 tool. Mean differences or standardized mean differences (SMDs) with 95% CIs were pooled using random-effects models with the Hartung-Knapp-Sidik-Jonkman adjustment. Heterogeneity was quantified using I² and τ², with 95% prediction intervals reported where estimable. Certainty of evidence was evaluated using the GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach. Results: A total of 22 randomized controlled trials involving 2294 participants were included. Telerehabilitation was associated with significant improvements in HRQoL (SMD 1.26, 95% CI 0.07-2.45; P=.04), dyspnea (SMD 1.95, 95% CI 0.60-3.31; P=.005), and functional aerobic capacity (6-minute walk test : mean difference 77.79 meters, 95% CI 30.44-125.14; P=.001). Fatigue was also significantly reduced (SMD 0.89, 95% CI 0.16-1.62; P=.02), but no significant effects were observed for lower-limb strength (sit-to-stand test: SMD 0.55, 95% CI –0.15 to 1.25; P=.12) or symptoms of anxiety and depression (SMD 0.12, 95% CI –0.13 to 0.38; P=.36). No significant differences between delivery modes were observed for HRQoL (P=.93), functional aerobic capacity (P=.29), or dyspnea (P=.16). A significant subgroup difference was observed for lower-limb strength (P=.02). Conclusions: Telerehabilitation may improve quality of life, aerobic capacity, and symptom burden in PCC and offer an accessible option for post-COVID rehabilitation, although the certainty of evidence was generally limited. Asynchronous delivery appears feasible for general reconditioning, whereas its benefit for lower-limb strength and the value of synchronous supervision remain uncertain. Clinical Trial: PROSPERO CRD42023490863; https://www.crd.york.ac.uk/PROSPERO/view/CRD42023490863

Barriers and Facilitators in the Implementation of Digital, Blended, and Analog Evidence-Based Osteoarthritis Care: International Qualitative Expert Interview Study

2026-10-23T14:15:20-04:00

Background: Evidence-based nonsurgical osteoarthritis (OA) care programs improve patient outcomes and may reduce the burden of disease. Despite their demonstrated effectiveness, their implementation into routine health care remains limited across health care systems. Understanding factors that facilitate or hinder implementation is essential to support broader adoption and long-term sustainability worldwide. Objective: This study aimed to explore factors that facilitate or hinder the implementation of evidence-based nonsurgical OA care programs. Specifically, it examined implementation across different health care settings, implementation phases, and analog, digital, and blended care models to identify transferable insights for implementation in different national and regional contexts. Methods: We conducted a qualitative expert interview study with 12 international experts involved in evidence-based OA care programs implemented across 6 countries. Semistructured interviews explored experiences with program development, implementation, and further development of analog, digital, and blended care models. Interview transcripts were analyzed using qualitative content analysis according to Kuckartz, using deductive and inductive category development. Results: Key barriers included limited funding and reimbursement structures, insufficient political support, organizational complexity, and legal or regulatory requirements. Important facilitators comprised strong leadership, interdisciplinary collaboration, structured communication, adaptation to local health care contexts, and the use of existing international program structures. Participants further emphasized that implementation requirements varied across the preparation, initial implementation, and consolidation phases, highlighting the need for phase-specific implementation strategies. Conclusions: Successful implementation of evidence-based nonsurgical OA care programs requires sustainable funding, stakeholder engagement, context-specific adaptation, and structured implementation planning across all implementation phases. Existing international program structures and collaboration may facilitate implementation across health care systems while reducing development efforts. Future implementation initiatives should consider phase-specific requirements and local contextual factors to improve long-term adoption and sustainability.

The SmartGlove BISINDO Wearable Sensor and Large Language Model–Based Sign Language–to-Speech System for Users With Communication Disorders: Design Science Research Study

2026-10-22T17:30:15-04:00

Background: In Indonesia, people with communication disorders predominantly communicate using Bahasa Isyarat Indonesia (BISINDO). Nevertheless, all current systems for BISINDO recognition generate fragmented, word-by-word output and lack natural language generation (NLG) capabilities, thereby fundamentally constraining their communicative utility. Objective: The aim of this study is to develop, validate, and field-evaluate SmartGlove BISINDO, an AI-assisted wearable sign language–to-speech system that embeds large language model (LLM)–powered NLG into a BISINDO translation pipeline for the first time. Methods: This study followed the design science research (DSR) framework, covering phases 1 through 5 (problem identification, solution objectives, prototype design and development, demonstration, and evaluation). Data were collected from 12 purposively selected informants (6 domain experts and 6 deaf education companions/teachers) across 3 Indonesian cities (Malang, Surabaya, and Makassar), using a needs-analysis questionnaire and an expert validation sheet, and analyzed through statistical triangulation of agreement percentage, content validity ratio (CVR), and Aiken V. The validated prototype was field-tested in Makassar and Malang with 15 participants with communication disorders and 4 observers or companions; it was evaluated by a disability expert in Surabaya. Results: The needs analysis yielded a grand mean of 4.77 (SD 0.51) across 4 clusters, with 89% mean agreement, indicating very high priority for all identified needs. The prototype’s 5-layer architecture integrates wearable sensors, on-device gesture classification, and LLM-based NLG. Expert validation showed an overall Aiken V of 0.94, with all 14 items exceeding the 0.75 validity cutoff and with 11/14 items (79%) rated highly valid (V>0.90). The grand mean CVR was 0.76 (SD 0.16), exceeding the Lawshe minimum threshold of 0.56 (<.05), with 3 items—gesture differentiation, AI contextual relevance, and settings configuration ease—falling below the threshold and indicating areas for refinement. During the testing phase, the system received a lower practicality score (mean 3.35, SD 0.16; 67%), particularly for clarity of speech output. However, observers and companions gave a significantly higher score of 88%, with a mean of 4.40 (SD 0.08). A disability expert also gave a score of 81% (with a mean value calculated across the 20 questionnaire items of 4.05, SD 0.60). Regarding language, the AI-generated output underwent three transformation processes: addition, deletion, and substitution. Of the 3, substitution is the most likely to alter the intended meaning. Conclusions: SmartGlove BISINDO is a wearable sign language–to-speech system that introduces the BISINDO research landscape to LLM-based NLG connect-and-translate approaches and demonstrates strong content validity. This study presents preliminary evidence toward a replicable model for interdisciplinary DSR-based prototype development that combines augmentative and alternative communication (AAC) theory, systemic functional linguistics, and design science methodology, intended as a reference for other underrepresented sign language contexts in Southeast Asia and beyond. Field evaluation further indicated favorable practicality overall, alongside a gap between expert-perceived feasibility and user-rated experience that will guide subsequent design refinement.