NHRI Intelligent Voice Interactive Platform for High-Quality Community Living Among Older Adults

Community-dwelling older adults in this single-arm interventional study use the NHRI-SmartCare intelligent voice platform at participating community senior centers, indigenous cultural health stations, or local housing communities over a 6-month intervention period. Following onboarding and training, participants engage in routine voice interactions for health and anti-fraud inquiries, read or listen to audio-read digital health newsletters, upload physiological metrics for self-health management, navigate nearby daily living and long-term care resources, and receive ongoing technical support from the NHRI research team. Comprehensive health-related quality of life, intrinsic capacity, cognitive, social participation, physical fitness, and platform usability assessments are administered at baseline and at 6-month follow-up.

The NHRI- Smart Voice Interaction Platform is an AI-driven, voice-interactive behavioral care platform customized for older adults to address health information-seeking barriers and enhance self-management. It features five core components:

1. Interactive Voice Consultation: Natural-language voice Q&A providing reliable health guidance (chronic diseases, nutrition, pain care, mental wellbeing) and anti-fraud counseling.
2. Bimodal Digital Newsletters: Health and wellness bulletins accessible via on-screen reading or on-demand spoken audio read-aloud.
3. Self-Monitoring: Digital tracking for uploading physiological metrics (blood pressure, blood glucose) to foster daily self-care.
4. Community Resource Navigation: Directory enabling users to search for nearby medical services, long-term care, and living essentials.
5. Technical Support: Routine assistance by the NHRI team to maintain digital engagement.

Health-related quality of life is assessed using the SF-36 Taiwan Version, evaluating 8 subscales: Physical Functioning (PF, 10 items, raw 10-30), Role-Physical (RP, 4 items, raw 4-8), Bodily Pain (BP, 2 items, raw 2-12), General Health (GH, 5 items, raw 5-25), Vitality (VT, 4 items, raw 4-24), Social Functioning (SF, 2 items, raw 2-10), Role-Emotional (RE, 3 items, raw 3-6), and Mental Health (MH, 5 items, raw 5-30). Per standard Taiwanese scoring guidelines, raw scores are linearly transformed into an identical 0 to 100 standardized scale: [(raw score - min score) / score range] * 100. The outcome reports changes in each standardized subscale score (Unit of Measure: points, range 0-100), where higher scores indicate superior functional health. Changes from baseline to 6 months are evaluated.

Self-care and daily health practices are assessed using the Health-Promoting Lifestyle Profile II (HPLP-II, Chinese version). The instrument evaluates 52 items across six dimensions: health responsibility, physical activity, nutrition, spiritual growth, interpersonal relations, and stress management, each rated on a 4-point Likert scale (1 = "never" to 4 = "routinely"). The outcome is reported as the overall mean composite score across all items (Unit of Measure: score on a scale of 1.0 to 4.0), where higher scores indicate more frequent adoption of health-promoting self-management behaviors. Changes from baseline to 6 months are evaluated.

Upper-body muscular strength and endurance are assessed using the 30-Second Arm Curl Test from the Senior Fitness Test protocol. Participants sit upright and complete as many biceps curls as possible within 30 seconds using their dominant hand (5-pound dumbbell for females, 8-pound for males). The outcome is reported as the total count of correctly completed repetitions (Unit of Measure: repetitions). For clinical interpretation, repetition counts are categorized against sex- and age-specific normative tiers (Excellent, Good, Average, Below Average, Poor). Higher counts indicate superior functional muscle strength. Changes from baseline to 6 months are evaluated.

Mobility and gait function are assessed using the 4-Meter Walk Test adapted from the Short Physical Performance Battery (SPPB) protocol. Participants walk a 4-meter course at their usual, comfortable pace. Walk performance is converted into a 4-point ordinal category reflecting functional mobility tiers: 3 = <3.5 seconds (fast gait speed, superior mobility), 2 = 3.5-4.9 seconds, 1 = 5.0-8.0 seconds, and 0 = >8.0 seconds (slow gait speed, mobility impairment). The outcome is reported as the change in this single ordinal category score (Unit of Measure: category on a scale of 0 to 3), where higher scores indicate superior functional mobility. Changes from baseline to 6 months are evaluated.

Lower-extremity functional strength and dynamic balance are measured by recording the maximum number of full stands completed from a straight-backed chair within 30 seconds, with arms folded across the chest. The outcome is reported as the total count of completed stands (Unit of Measure: stands). A higher number indicates superior lower-body muscular strength, endurance, and functional mobility. Changes from baseline to 6 months are evaluated.

Nutritional status is evaluated using the full 18-item Mini Nutritional Assessment (MNA). Assessment follows a validated two-step procedure: Step 1 Screening (items A-F, max 14 points: 12-14 indicates normal nutritional status; 8-11 indicates risk of malnutrition; 0-7 indicates malnutrition). Participants scoring <=11 points proceed to Step 2 Assessment (items G-R, max 16 points) to derive the full Malnutrition Indicator Score (sum of screening and assessment, max 30 points). The outcome is reported as the continuous total MNA score (Unit of Measure: points, range 0.0 to 30.0), classified as: 24.0-30.0 (normal status), 17.0-23.5 (at risk of malnutrition), and <17.0 (malnourished). Higher scores indicate better nutritional status. Changes from baseline to 6 months are evaluated.

Subjective feelings of loneliness and social isolation are assessed using the 20-item UCLA Loneliness Scale (Version 3). Items are rated on a 4-point Likert scale from 1 ("never") to 4 ("often"). The outcome is reported as the total sum across all items (Unit of Measure: points, range 20 to 80), categorized as: 20-34 (low degree of loneliness), 35-49 (moderate degree of loneliness), 50-64 (moderately high degree of loneliness), and 65-80 (high degree of loneliness). Lower scores reflect reduced feelings of loneliness and stronger social connectedness. Changes from baseline to 6 months are evaluated.

Perceived psychological stress over the past month is measured using the 14-item Perceived Stress Scale (PSS-14). Positively stated items (4, 5, 6, 7, 9, 10, 13) are reverse-scored from 4 ("never") to 0 ("very often"), while remaining items (1, 2, 3, 8, 11, 12, 14) are scored from 0 ("never") to 4 ("very often"). The outcome is reported as the total composite score (Unit of Measure: points, range 0 to 56), categorized as: 0-28 (normal), 29-42 (elevated stress), and 43-56 (severe stress requiring assistance). Higher scores indicate greater perceived psychological stress. Changes from baseline to 6 months are evaluated.

Depressive symptoms over the past week are evaluated using the 15-item Geriatric Depression Scale (GDS-15) in a dichotomous format (Yes/No). Positively worded items (1, 5, 7, 11, 13) are scored 1 point for "No" and 0 for "Yes", while depressive items (2, 3, 4, 6, 8, 9, 10, 12, 14, 15) are scored 1 point for "Yes" and 0 for "No". The outcome is reported as the single continuous total score (Unit of Measure: points, range 0 to 15), categorized as: 0-6 (well-adapted, normal status), 7-10 (moderate emotional distress), and 11-15 (severe emotional distress recommending psychiatric referral). Lower scores reflect fewer depressive symptoms and superior affective wellbeing. Changes from baseline to 6 months are evaluated.

Global cognitive screening is evaluated using the TDS Brain Health Test (BHT), assessing four domains: Orientation (4 points), Word Registration (5 points), Category Verbal Fluency (animal naming within 60 seconds: >=9 words = 2 points, 5-8 words = 1 point, <5 words = 0 points), and Word Recall (5 points). The outcome is reported as the total composite screening score (Unit of Measure: points, range 0 to 16), where scores <=9 indicate suspected mild dementia or cognitive impairment, and scores >=10 reflect normal cognitive function. Higher scores represent superior global cognitive performance. Changes from baseline to 6 months are evaluated.

Intrinsic capacity is assessed using the World Health Organization Integrated Care for Older People (WHO ICOPE) screening tool, covering six domains: cognitive decline, limited mobility, malnutrition, visual impairment, hearing loss, and depressive symptoms. The outcome is reported as the total count of abnormal/impaired domains (Unit of Measure: domains, range 0 to 6). A lower count indicates better-preserved intrinsic capacity and overall functional health. Changes from baseline to 6 months are evaluated.

Global cognitive performance is measured using the Taiwan Version of the Montreal Cognitive Assessment (MoCA). It assesses 7 domains: visuospatial/executive, naming, attention, language, abstraction, delayed recall, and orientation. The outcome is reported as the continuous total score, adjusted for education (1 extra point added for <=12 years of education; Unit of Measure: points, range 0 to 30). A score >=26 denotes normal cognition. Higher scores represent superior cognitive function. Changes from baseline to 6 months are evaluated.

Activity diversity is evaluated using the 24-item PM-3D4D across Social (6 items), Community (12 items), and Productivity (6 items) domains. Participants indicate whether they performed each activity during the past 3 months (dichotomous: 0 = No, 1 = Yes). The outcome is reported as the overall Diversity Score, defined as the proportion of engaged activities (number of affirmative responses divided by 24 total items; Unit of Measure: ratio, range 0.00 to 1.00). Higher ratios indicate greater activity variety. Changes from baseline to 6 months are evaluated.

Activity participation frequency over the past 3 months is assessed using the 24-item PM-3D4D across Social (6 items, rated 0-6), Community (12 items, rated 0-6), and Productivity (6 items, rated 0-4) domains. Consistent with the original scoring manual metrics, the outcome evaluates changes in domain-specific summated frequency scores: Social domain (Unit of Measure: points, range 0 to 36), Community domain (Unit of Measure: points, range 0 to 72), and Productivity domain (Unit of Measure: points, range 0 to 24). Higher scores reflect more frequent social, community, and productive participation. Changes from baseline to 6 months are evaluated.

Participation difficulty is assessed using the Traditional Chinese PM-3D4D (24 items across Social [6 items], Community [12 items], and Productivity [6 items] domains). For each performed activity over the past 3 months, perceived difficulty is rated on a 4-point scale (1 = "Very difficult" to 4 = "Not difficult at all"). Raw subscale difficulty scores are converted into standardized 0 to 100 scores via Rasch-based conversion tables and averaged across domains to yield a composite Total Difficulty Score (Unit of Measure: score on a scale of 0 to 100). Higher scores reflect greater ease and less difficulty in participation. Changes from baseline to 6 months are evaluated.

Participants' acceptance of the smart voice platform is evaluated using an adapted 11-item Technology Acceptance Model (TAM) questionnaire. Each item is scored on a 7-point Likert scale ranging from 1 ("strongly disagree") to 7 ("strongly agree"). The outcome is reported as the total sum score across all 11 items (Unit of Measure: points, range 11 to 77), evaluating perceived usefulness, perceived ease of use, and behavioral intention to use the platform. Higher scores represent greater technology acceptance. Changes from baseline to 6 months are evaluated.

Platform usability and user satisfaction are evaluated using the 10-item System Usability Scale (SUS). Each item is rated on a 5-point Likert scale (1 = "strongly disagree" to 5 = "strongly agree"). Raw item scores are converted according to the standard Brooke scoring algorithm into a composite score from 0 to 100 (Unit of Measure: points, range 0 to 100). Higher scores indicate superior platform usability (scores >=68 represent acceptable usability). Changes from baseline to 6 months are evaluated.

Adoption of daily digital services is assessed across 7 domains: online information search, mobile payments, online finance/stocks, social messaging, digital healthcare/telehealth, online entertainment, and digital learning. Each domain is scored dichotomously (0 = No, 1 = Yes). The outcome is reported as the total count of regularly utilized digital services (Unit of Measure: counts, range 0 to 7). Higher counts indicate broader digital autonomy in community living. Changes from baseline to 6 months are evaluated.

Perceived informal emotional and social support is assessed through 3 core items: willingness of others to listen (1-5), perceived care from family/friends (1-5), and satisfaction with emotional support received (1-5). The outcome is reported as the mean composite score across the three items (Unit of Measure: score on a scale of 1.0 to 5.0). Higher scores indicate stronger perceived informal social support. Changes from baseline to 6 months are evaluated.

Assessed using the formal educational attainment classification. Educational levels are categorized as: 1 = Elementary school or below (≦6 years), 2 = Junior high school (7-9 years), 3 = Senior high/vocational school (10-12 years), 4 = Junior college, 5 = University (bachelor's degree), and 6 = Master's degree or above. The outcome is measured by the formal educational level (categorical tier 1 to 6). Higher scores represent higher educational attainment and formal cognitive reserve.

Multimorbidity burden is assessed using a standardized checklist of 13 physician-diagnosed chronic conditions (e.g., hypertension, diabetes, hyperlipidemia, cardiovascular disease, stroke, arthritis). Each condition is scored dichotomously (0 = No, 1 = Yes). The outcome is reported as the total count of co-existing chronic conditions (Unit of Measure: counts, range 0 to 13). Changes from baseline to 6 months are evaluated.

Self-rated general health status is assessed using a single-item standard 5-point scale: 1 = "Very poor", 2 = "Poor", 3 = "Fair", 4 = "Good", 5 = "Excellent". The outcome is reported as the numeric rating (Unit of Measure: score on a scale of 1 to 5), where higher scores reflect superior subjective perception of health and wellbeing. Changes from baseline to 6 months are evaluated.

Perceived community environmental quality of life is evaluated using 5 items adapted from the Environment domain of the WHOQOL-BREF (Taiwan Version), addressing information access, leisure, physical living environment, healthcare access, and transportation. Each item is scored on a 5-point Likert scale (1 = "Very dissatisfied/Inconvenient" to 5 = "Very satisfied/Convenient"). The outcome is reported as the raw summated score across the 5 items (Unit of Measure: points, range 5 to 25). Higher scores indicate greater satisfaction with community environment and resource accessibility. Changes from baseline to 6 months are evaluated.

Routine self-monitoring adherence for blood pressure is assessed. Measurement frequency is scored on a 4-point ordinal scale: 0 = Never/rarely, 1 = At least once a year, 2 = At least monthly, 3 = 1-2 times weekly. The outcome is reported as the ordinal frequency score (Unit of Measure: score on a scale of 0 to 3). Higher scores indicate more proactive cardiovascular self-monitoring habits. Changes from baseline to 6 months are evaluated.

Routine self-monitoring adherence for blood glucose is assessed. Measurement frequency is scored on a 4-point ordinal scale: 0 = Never/rarely, 1 = At least once a year, 2 = At least monthly, 3 = 1-2 times weekly. The outcome is reported as the ordinal frequency score (Unit of Measure: score on a scale of 0 to 3). Higher scores indicate more proactive metabolic self-monitoring habits. Changes from baseline to 6 months are evaluated.

Habitual physical activity duration over the past month is assessed by multiplying weekly exercise frequency by daily duration across activities such as walking, cycling, swimming, hiking, ball sports, dancing, or qigong. Weekly accumulated exercise time is classified into a 4-point ordinal category based on public health activity recommendations: 3 = >=360 minutes/week, 2 = 180 to <360 minutes/week, 1 = 60 to <180 minutes/week, and 0 = <60 minutes/week. The outcome is reported as the change in this ordinal category score (Unit of Measure: category on a scale of 0 to 3), where higher scores reflect greater habitual physical activity. Changes from baseline to 6 months are evaluated.