JMIR Aging最新文献

Background: Falls are one of the leading causes of injury or death among older adults. Falls occurring in individuals during hospitalization, as an adverse event, are a key concern for health care institutions. Identifying older adults at high risk of falls in clinical settings enables early interventions, thereby reducing the incidence of falls.

Objective: This study aims to develop and validate machine learning models to predict the risk of falls among hospitalized older adults.

Methods: This study retrospectively analyzed data from a tertiary general hospital in China, including 342 older adults who experienced falls and 684 randomly matched nonfallers, between January 2018 and December 2024, encompassing demographic information, comorbidities, laboratory parameters, and medication use, among other variables. The dataset was randomly split into training and testing sets in a 7:3 ratio. Predictors were selected from the training set using stepwise regression, least absolute shrinkage and selection operator, and random forest-recursive feature elimination. Seven machine learning algorithms were employed to develop predictive models in the training set, and their performance was compared in the testing set. The optimal model was interpreted using Shapley Additive Explanations (SHAP).

Results: The gradient boosting machine model demonstrated the best predictive performance (C-index 0.744, 95% CI 0.688-0.799). The 8 most important variables associated with fall risk were dizziness, epilepsy, fall history within the past 3 months, use of walking assistance, emergency admission, Morse Fall Scale scores, modified Barthel Index scores, and the number of indwelling catheters. The model was interpreted using SHAP to enhance the clinical utility of the predictive model.

Conclusions: The gradient boosting machine model was identified as the optimal predictive model. The SHAP method enhanced its integration into clinical workflows.

Background: The theory of complexity in aging indicates that the complexity of sensor-derived physiological and behavioral signals reflects an older adult's adaptive capacity and, in turn, their frailty. Smart homes with ambient sensors offer a unique opportunity to longitudinally explore the complexity of older adults' indoor movement in a real-world setting. Here, we introduce a computational method to estimate behavior complexity from sensor data. We further conduct a multiple-methods case series to explore the relationship between entropy-measured smart home data complexity and older adult frailty.

Objective: This study aims to explore the relationship between entropy-measured ambient sensor data complexity and frailty in independent community-dwelling older adults.

Methods: The nature of older adults' indoor movement complexity is measured by quantifying the entropy of smart home data. Overall, 11 cases with persons aged 65 years and older were drawn from an ongoing smart home study to illustrate the method. We assessed weekly frailty for these cases using the Clinical Frailty Scale. For corresponding time ranges, we measured the complexity of smart home data using a fixed-width sliding window and an entropy-based complexity index (Rényi Complexity Index) built on a Universal Sequence Map (USM-Rényi). Descriptive statistics and graphical analysis were used to describe intraindividual frailty and sensor complexity change.

Results: The complexity of sensor-observed indoor movement does change over time in older adults as quantified by the computational method. In some individuals, these changes track with health transitions and frailty progression. The trends and monotonicity of complexity trajectories varied between cases. Overall, 3 of the cases demonstrated a negative association between frailty and complexity, while the association was not as clear for the other cases.

Conclusions: The complexity of older adults' smart home data is highly diverse. Changes in health and frailty influence indoor movement complexity. Although the findings suggest a relationship between frailty and complexity, confounding factors, such as home layout, visitors, external events, and technology disruptions, may influence sensor signals.

Background: Mild cognitive impairment (MCI) is an intermediate state between normal aging and dementia, characterized by subjective cognitive decline and objective memory impairment. Cognitive training has consistently shown short-term benefits for individuals with MCI, but evidence on the long-term effectiveness is extremely limited. Given the progressive nature of MCI and the need for sustainable strategies to delay cognitive decline, research on the long-term impact of cognitive training is necessary and timely. Mobile-based platforms offer a promising solution by enhancing accessibility and adherence, but their durability of effect over extended periods remains underexplored.

Objective: This study aimed to evaluate the long-term effects of a mobile-based cognitive training app on the cognitive function of older adults with MCI.

Methods: In total, 28 older adults with MCI used Cogthera, a mobile cognitive training app based on metamemory training. Participants completed 2 training sessions daily for 3 months, and 9 (32%) continued for an additional 12 months. Cognitive function and quality of life were assessed using the Alzheimer's Disease Assessment Scale-Cognitive Subscale 14 and EQ-5D-5L.

Results: Cognitive function improved over 15 months, as measured by Alzheimer's Disease Assessment Scale-Cognitive Subscale (F_2,35.56=7.08; P=.003). EQ-5D-5L scores increased at 3 months but did not show sustained change at 15 months (F_2,42.14=3.40; P=.04). Greater cognitive improvements were associated with younger age, higher functional status, and lower baseline cognitive function.

Conclusions: This study showed that long-term use of a mobile-based metamemory cognitive training app was associated with cognitive improvements over 15 months. Although limited by the small sample size and the absence of a control group, these findings suggest potential for mobile cognitive training as a sustainable intervention that warrants validation in larger trials.

Background: More than half of people with HIV are now older than 50 years, and they face an approximately 60% higher risk of developing dementia compared with the general population. In recent years, the application of artificial intelligence, particularly machine learning, combined with the growing availability of large datasets, has opened new avenues for developing prediction models that improve dementia detection, monitoring, and management.

Objective: This systematic review aimed to synthesize the existing literature on the application of machine learning in dementia research among older people with HIV and identify directions for future research.

Methods: A comprehensive search was conducted in PubMed, CINAHL, and Embase in September 2024, limited to studies published within the past 10 years. Eligible articles included original research involving people with HIV applying at least 1 machine learning technique and reporting dementia-related outcomes.

Results: The search yielded 721 articles, of which 26 (3.6%) met the inclusion criteria. Most studies were retrospective and conducted in the United States (n=14, 53.8%), primarily focusing on neurocognitive impairment and HIV-associated neurocognitive disorders. Supervised machine learning techniques were most frequently used and demonstrated strong predictive performance. Common methodological challenges included small sample sizes, lack of external validation, limited participant diversity, and concerns about biological interpretability and generalizability.

Conclusions: Machine learning research on dementia among older people with HIV primarily targets HIV-associated neurocognitive disorders, with limited exploration of age-related neurodegenerative diseases such as Alzheimer disease and related dementias. The absence of longitudinal studies and external validation remains a key limitation. Future research should broaden the focus to all-cause dementia beyond HIV-specific conditions; apply advanced machine learning methods; and leverage large-scale longitudinal, multimodal datasets. Strengthening methodological rigor and enhancing real-world applications will be critical to improving early detection and effective management of cognitive health in this unique aging population.

Background: Cardiovascular disease (CVD) is the main cause of death in middle-aged and older people in China. The interplay between sarcopenia and insulin resistance (IR) in driving CVD risk has not been fully understood, particularly regarding sarcopenia severity and IR heterogeneity.

Objective: This study aimed to investigate the relationship between IR and sarcopenia and the risk of new-onset CVD.

Methods: Using data from the China Health and Retirement Longitudinal Study (CHARLS). Cox proportional hazards models were used to assess associations of sarcopenia status (nonsarcopenia, possible sarcopenia, sarcopenia, and severe sarcopenia) and 6 IR indices (triglyceride-glucose, TyG; TyG-BMI; TyG-waist circumference; TyG-waist-to-height ratio; triglyceride/high-density lipoprotein cholesterol, TG/HDL-C; and metabolic score for insulin resistance, METS-IR) with incident CVD. Additive and multiplicative interaction analyses and subgroup analyses by age and sex were performed. Receiver operating characteristic analysis was used to determine clinically relevant cutoffs.

Results: In this study, during a median 9-year follow-up, we included 5514 middle- and older-aged (≥45 y) residents, of whom 550 presented with CVD incidence. Participants with possible sarcopenia and high IR exhibited 1.24-1.85-fold higher CVD risk versus nonsarcopenia and low-IR counterparts (P<.05) after adjustment for potential confounders. While TyG-BMI and TyG-waist circumference were the strongest independent predictors, formal interaction analysis revealed that the TG/HDL-C ratio and METS-IR demonstrated the most consistent synergistic effects with possible sarcopenia (relative excess risk due to interaction=0.139 and 0.074, respectively). In subgroups of different ages and sexes, the combination of IR and sarcopenia is associated with the highest risk of CVD. Receiver operating characteristic analysis provided clinically applicable cutoffs for these indices, including TG/HDL-C ≥2.09 and METS-IR ≥34.26.

Conclusions: We found that IR and sarcopenia, especially early-stage sarcopenia, synergistically increase the incidence of CVD in older adults. These findings advocate for dual-targeted CVD interventions (muscle preservation and IR mitigation) in aging societies, particularly during the transitional phase of possible sarcopenia.

Background: Artificial intelligence (AI) has demonstrated superior diagnostic accuracy compared with medical practitioners, highlighting its growing importance in health care. SMART-Pred (Shiny Multi-Algorithm R Tool for Predictive Modeling) is an innovative AI-based application for Alzheimer disease (AD) prediction using handwriting analysis.

Objective: This study aimed to develop and evaluate a noninvasive, cost-effective AI tool for early AD detection, addressing the need for accessible and accurate screening methods.

Methods: The study used principal component analysis for dimensionality reduction of handwriting data, followed by training and evaluation of 10 diverse AI models, including logistic regression, naïve Bayes, random forest, adaptive boosting, support vector machine, and neural network. Model performance was assessed using accuracy, sensitivity, precision, specificity, F1-score, and area under the curve (AUC) metrics. The DARWIN (Diagnosis Alzheimer With Handwriting) dataset, comprising handwriting samples from 174 participants (89 patients with AD and 85 healthy controls), was used for validation and testing.

Results: The neural network classifier achieved an accuracy of 91% (95% CI 0.79-0.97) and an AUC of 94% on the test set after identifying the most significant features for AD prediction. These performance results surpass those of current clinical diagnostic tools, which typically achieve around 81% accuracy. SMART-Pred's performance aligns with recent AI advancements in AD prediction, such as Cambridge scientists' AI tool achieving 82% accuracy in identifying AD progression within 3 years, using cognitive tests and magnetic resonance imaging scans. The variables "air_time" and "paper_time" consistently emerged as critical predictors for AD across all 10 AI models, highlighting their potential importance in early detection and risk assessment. To augment transparency and interpretability, we incorporated the principles of explainable AI, specifically using Shapley Additive Explanations, a state-of-the-art method to emphasize the features responsible for our model's efficacy.

Conclusions: SMART-Pred offers noninvasive, cost-effective, and efficient AD prediction, demonstrating the transformative potential of AI in health care. While clinical validation is necessary to confirm the practical applicability of the identified key variables, the findings of this study contribute to the growing body of research on AI-assisted AD diagnosis and may lead to improved patient outcomes through early detection and intervention.

Background: Older adults make up the largest proportion of nonusers of the internet. With the increasing digitalization of services, it is important to identify what interventions are effective at reducing digital exclusion in older adults.

Objective: We aimed to identify what evidence exists on the effectiveness of interventions to address digital exclusion in older adults.

Methods: This rapid review assessed the effectiveness of interventions to address digital exclusion in older adults aged 60 years or older. Searches were conducted in November 2023 across a range of databases and used supplementary search methods. Searches were limited to comparative studies published from 2018 onward in English. Data were analyzed using a narrative synthesis approach.

Results: A total of 21 studies were included that aimed to increase a range of digital literacy skills. Sample sizes ranged from 5 to 381. Intervention approaches varied considerably and were often multicomponent and undertaken in a variety of settings. There is evidence to suggest that a range of interventions can reduce physical, personal, and perceptual barriers and improve older adults' skills, knowledge, digital literacy, and perceived self-efficacy, reduce technophobia, and increase use of technology. Importantly, findings indicated improvements among a range of subpopulations, including those living in rural areas, at risk of social isolation, who are homebound, of lower socioeconomic groups, and individuals with visual impairment. To achieve improved and sustained digital inclusion in older adults, evidence suggests it may be important to ensure structural barriers, such as access to the internet and affordability of devices, are removed. However, all studies contained methodological limitations and may not be adequately powered to determine effectiveness.

Conclusions: The evidence shows the potential benefits of interventions aimed at improving a range of digital skills and increasing technology use in older adults, which could help to address digital exclusion. The findings of this rapid review can inform the development and delivery of future interventions. However, it is important to consider the context in which the included interventions were used and the lack of certainty of the findings. This review also identified a lack of high-quality evidence, as all studies identified contained methodological limitations and may not have been adequately powered to determine effectiveness. In addition, consideration should also be given to those who do not wish to engage with the online world to ensure they are not left behind.

Background: Dementia presents substantial challenges for informal caregivers. A gap remains in technology-driven personalized support tailored to caregivers' needs.

Objective: This study aimed to develop a theory-driven, multicomponent mobile app specifically designed for caregivers of individuals with dementia and test its usability among end users.

Methods: We developed CareBuddy, a mobile care ecosystem based on the stress process model and user-centered design. The app includes personalized assessments and tailored solutions, an artificial intelligence-driven chatbot, GPS-based location monitoring, peer support, a helpline, telemedicine, health care provider integration, and caregiver self-care resources. Development was informed by interviews with caregivers and stakeholders, followed by a 2-phase pilot test involving 18 and 10 participants, respectively, to assess usability and acceptability.

Results: In phase 1, the mean system usability scores increased from 65.4 (SD 11.8) in round 1 to 73.8 (SD 15.9) in round 3, exceeding the benchmark of 68. In phase 2, caregivers rated the app highly, with an overall mean score of 95.4 (SD 8.5) on the Mobile Health App Usability Questionnaire. The domains of ease of use (mean 24.1, SD 2.9), user interface and satisfaction (mean 40.3, SD 3.4), and usefulness (mean 31, SD 3.9) received high Mobile Health App Usability Questionnaire ratings. Participants valued the content focused on dementia management and caregiver well-being. Caregivers appreciated the interactive features: social networking portal, service directory, and conversational large language model. Feedback highlighted areas for improvement, including reducing textual overload and addressing navigational challenges.

Conclusions: CareBuddy offers a multifaceted digital solution for dementia caregivers, with high usability and satisfaction. An ongoing trial is evaluating the app's effectiveness in improving caregiver outcomes.

Background: The health and economic burden of dementia has led the World Health Organization to recognize it as a public health priority. Although there currently does not exist a cure for dementia, there are multiple interventions aimed at preventing the risk of dementia and improving the quality of life of people with dementia. Voice assistants (VAs), particularly those using large language models (LLMs), have emerged as promising tools to deliver these interventions to older adults due to their accessible and natural interface.

Objective: This pilot study aimed to evaluate the technical feasibility (ie, functional performance and usability) and user acceptance of the embodied rule-based and LLM VA GRACE, as well as the perceived strength of the collaborative relationship or working alliance, between GRACE and healthy older adults during the delivery of cognitive stimulation interventions.

Methods: A pilot study was conducted with 21 healthy German-speaking adults aged 60 years and older. Participants interacted with GRACE in a laboratory setting for 10-15 minutes. The interaction involved a structured cognitive stimulation session using rule-based and LLM components. Data were collected using pre- and postinteraction questionnaires and semistructured interviews. Quantitative analysis included descriptive statistics and Wilcoxon signed rank tests. Qualitative data were analyzed thematically.

Results: Participants rated GRACE positively, with statistically significant scores above neutral (P<.001 for perceived ease of use, usefulness, enjoyment, and working alliance; P=.009 for perceived control; and P=.009 for intention to continue interacting). Thematic analysis revealed that GRACE was perceived as easy to understand and unambiguous, friendly, and supportive, with intervention components viewed as enjoyable and appropriately challenging. Areas for improvement included personalization, response delays, and voice quality.

Conclusions: The results suggest that embodied rule-based and LLM VAs like GRACE are feasible and well-received tools for delivering cognitive interventions to older adults. Future iterations will incorporate feedback and extend testing to individuals at risk for dementia.

Background: Baseline sarcopenia has been linked to cognitive decline in older adults; however, the impact of longitudinal changes in sarcopenia status on cognitive trajectories remains unclear.

Objective: This aims to examine the association between 2-year transitions in sarcopenia status and subsequent 5-year cognitive trajectories among middle-aged and older adults in China.

Methods: We analyzed data from 8189 participants (median age 58, IQR y; n=432952.9% female) in the China Health and Retirement Longitudinal Study. Sarcopenia status was determined in 2011 and 2013 according to the 2019 Asian Working Group for Sarcopenia criteria, and participants were classified into 7 transition groups based on status changes. Cognitive function was assessed from 2013 to 2018 using standardized z scores for executive function and episodic memory. Linear mixed-effects models were applied to evaluate associations between sarcopenia transitions and cognitive trajectories, adjusting for demographic, lifestyle, and health-related covariates.

Results: Progression from a nonsarcopenic state was associated with greater cognitive decline compared to stable nonsarcopenia (β=-0.016, 95% CI -0.026 to -0.007; P<.001), with greater decline observed among those progressing from possible sarcopenia to sarcopenia (β=-0.027, 95% CI -0.054 to -0.001; P=.04). In contrast, regression from possible sarcopenia was associated with improved cognitive performance (β=0.028, 95% CI 0.015-0.041; P<.001). No significant improvement was observed among individuals regressing from established sarcopenia. Subgroup analyses showed consistent benefits of regression from possible sarcopenia across sex, age, residence, and education groups, except among urban residents (P=.05).

Conclusions: Progression in sarcopenia status was independently associated with accelerated cognitive decline, whereas regression from possible sarcopenia predicted cognitive benefit. These findings highlight possible sarcopenia as a clinically actionable and potentially reversible stage and underscore the importance of early identification and intervention to preserve cognitive health in aging populations.