BackgroundDepression is a common complication after a stroke that may lead to increased disability and decreased quality of life. The objective of this study was to develop and validate an interpretable predictive model to assess the risk of depression in stroke patients using machine learning (ML) methods.MethodsThis study included 1143 stroke patients from the NHANES database between 2005 and 2020. First, risk factors for depression in stroke patients were determined by univariate and multivariate logistic regression analysis. Next, five machine learning algorithms were used to construct predictive models, and several evaluation metrics (including area under the curve (AUC)) were used to compare the predictive performance of the models. In addition, the SHAP (Shapley Additive Explanations) method was used to rank the importance of features and to interpret the final model.ResultsWe screened seven features to construct a predictive model. Among the 5 machine learning models, the XGBoost (extreme gradient boosting) model showed the best discriminative ability, with an AUC of the ROC (receiver operating characteristic curve) in the test set of 0.746 and an accuracy of 0.834. In addition, the prediction results of the XGBoost model were interpreted in detail using the SHAP algorithm. We also developed a web-based calculator that provides a convenient tool for predicting the risk of depression in stroke patients at the following link: https://prediction-model-for-depression.streamlit.app.ConclusionsOur interpretable machine learning model serves as an auxiliary tool for clinical judgment, aimed at early and effective identification of depression risk in stroke patients.

ObjectiveThe relationship between sarcopenic obesity and masticatory function is poorly understood. This study aims to explore this association in community-dwelling individuals aged 75 years or older.MethodsThis study analyzed data from 236 community-dwelling adults aged 75 years or older. Masticatory function was assessed using spectrophotometric measurement of gum color differences before and after chewing color-changeable gum (ΔE*ab). Participants were categorized into tertiles of masticatory function based on their ΔE*ab values. The tertiles were defined as low, intermediate, and high. Sarcopenic obesity was assessed using the Consensus statement of the Japanese Working Group on Sarcopenic Obesity. Bayesian multinomial logistic regression was employed to examine the relationship between masticatory function and sarcopenic obesity.ResultsThe prevalence rates for obesity, sarcopenia, and sarcopenic obesity were 15.3%, 24.2%, and 9.7%, respectively. After adjusting for covariates, participants with high masticatory function had a significantly lower posterior estimate of sarcopenic obesity (posterior estimate: −1.83 [95% credible interval: −3.66, −0.22]) and sarcopenia (posterior estimate: −1.97 [95% credible interval: −3.37, −0.72]) compared with participants with low masticatory function. However, no significant associations were observed between masticatory function and obesity.ConclusionsThese findings suggest that high masticatory function is associated with a significantly lower prevalence of sarcopenic obesity in older adults.

BackgroundWith increasing age, individuals are more likely to experience physical disabilities, functional declines, and mobility limitations. Wearable robots or exoskeletons are relatively new technologies that can help address these issues, reduce healthcare costs, and support home healthcare, decreasing the burden of chronic disease. The purpose of this study was to investigate the usability of Bot Fit after task-specific physical activities and functional gait training, as well as to examine the effects of a wearable hip exoskeleton, Bot Fit, on gait, physical function, and muscle strength in older adults living in residential care facilities.MethodsA total of 32 older adults living in residential care facilities were included in this uncontrolled study. All participants performed eight weeks of task-specific physical activities and functional gait training using Bot Fit, with three exercise sessions per week (24 sessions in total). They were assessed at three time points: pre-test (baseline, T0), mid-test (after the 12 exercise sessions, T1), and post-test (after the last exercise session, T2). Each assessment evaluated functional outcomes (10-m walk test [10MWT], timed up-and-go [TUG], 6-min walk test [6MWT], Berg balance scale [BBS], four-square step test [FSST], and geriatric depression scale-short form [GDS-SF]), as well as muscle strength of the lower extremities. After the post-test, the participants completed a questionnaire to evaluate Bot Fit usability.ResultsA significant improvement was observed in all physical assessments, including the 10MWT, TUG, 6MWT, BBS, and FSST, from T0 to T2. It is noteworthy that 10MWT, TUG, and BBS also changed significantly from T0 to T1 and from T1 to T2. Muscle strength in hip flexion, hip extension, knee flexion, knee extension, ankle dorsiflexion, and ankle plantar flexion all improved significantly from T0 to T2, with knee flexion, knee extension, ankle dorsiflexion, and ankle plantar flexion showing significant improvements at all time points. Additionally, on the usability questionnaire, most participants provided positive feedback about their experience with Bot Fit.ConclusionThe findings of this study suggest that task-specific physical activity and functional gait training with Bot Fit have several key advantages for improving gait, physical function, and muscle strength in older adults living in residential care facilities. The findings support the application of Bot Fit to physical activity and functional gait training to improve age-related declines in physical function and muscle strength and to provide important insights into future robot-assisted exercise devices.Trial registrationURL: https://register.clinicaltrials.gov/. Unique identifier: NCT04610190 (10/26/2020).

BackgroundThis study investigates the association between elder abuse, pain with depression among older adults in India, with a focus on the interacting effect of gender. Elder abuse is a growing public health concern globally, and understanding its connection with pain and depression is crucial for prevention and intervention strategies, particularly in vulnerable demographic groups.MethodsData were drawn from the nationally representative Longitudinal Ageing Study in India (LASI) survey conducted in 2017-18 with the total sample size of 73,396. Study sample based on individuals aged 60 years and above, consisted 31,902 older adults. This study combines two binary variables pain and depression symptoms into a composite binary variable Pain with depression (Yes/No). Pain was assessed by asking question to the participants whether they are often troubled with pain. Depression was evaluated using the Centre for Epidemiological Studies depression Scale known as (CES-D-10), using four categories of scale options. A range from 0 to 10 of composite score is obtained and individual who score more than 4 were taken as depressed. Logistic regression models and Chi-square test of significance were used to analyse the relationship between elder abuse and pain with depression, while controlling for socio-demographic, functional and behavioural factors. Interaction effects of gender were examined to assess differential abuse risk between older male and female.ResultsThe analysis revealed that 5.2% of older adults reported experiencing abuse, with a higher prevalence among female. Older adults with pain and depression were significantly more likely to face abuse, with female showing consistently higher odds of abuse compared to male. Specific groups, such as those aged 75 and above, unmarried, uneducated and living in rural areas were at greater risk.ConclusionThe study highlights the strong association between elder abuse, pain with depression, especially among older female. These findings underscore the need for targeted public health interventions among vulnerable groups such as older female, and future research to explore cross- national dynamics and underlying risk factors.

BackgroundOrganophosphate esters (OPEs) are widely used as an alternative to the brominated flame retardant polybrominated diphenyl ethers. The effects of OPEs on the cognitive abilities of older adults remain unclear.MethodsA cross-sectional study was conducted using data from the National Health and Nutrition Examination Survey 2011–2014. Cognitive function was assessed using the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) word learning test, the CERAD word recall test, the Animal Fluency Test (AFT), and the Digit Symbol Substitution Test (DSST). OPE metabolites with detection rates above 50% were included in the study. Weighted multiple linear regression, weighted quantile sum (WQS) regression, and Bayesian kernel machine regression (BKMR) models were used to examine the effects of individual and mixed exposures to OPE metabolites on cognitive function.ResultsA total of 762 older adults were included. The weighted linear regression model revealed a positive association between Ln DPHP, Ln BDCPP, and Ln BCPP and the DSST score, while a negative association was observed between Ln DBUP and the DSST score. In the positive WQS model, the index was correlated with DSST score (β = 2.65, 95% CI: 0.40 ~ 4.90, P = 0.02), with DPHP having the highest weight. The results of BKMR analysis indicated a borderline statistical significance in the increase of DSST score when the mixture of OPEs is set to a specific 90th percentile compared to all mixture concentrations set to the median.ConclusionsOverall exposure to OPE metabolites are associated with improved cognitive function in older adults in the United States. Further prospective studies with large sample sizes are needed to confirm these results.

BackgroundStudies have shown that palliative care delivered to people living with dementia (PLWD) in nursing homes (NHs) improves care quality and reduces potentially burdensome treatments. However, access to palliative care services in NHs is uncommon. Telehealth may extend the reach of specialty palliative care consultation, yet strategies for feasible and acceptable NH implementation remain unknown. During implementation of an embedded pragmatic pilot clinical trial for PLWD, we aimed to describe key informants’ perceptions of a NH telehealth palliative care intervention.MethodsGuided by the Practical Implementation Sustainability Model (PRISM), we engaged key informants in 30–60-minute focus groups and individual semi-structured interviews to understand barriers and facilitators to implementation of a NH telehealth palliative care intervention in one NH. Interview prompts addressed contextual factors that influenced outcomes. Interviews were conducted and recorded via videoconference, transcribed, and analyzed using the Reach, Effectiveness, Adoption, Implementation, and Maintenance (RE-AIM) framework.ResultsParticipants (n = 14) included NH administrators and other leaders, palliative care providers, telehealth representatives, dementia advocates, a care partner, and a PLWD. Identified barriers to implementation included stigma surrounding dementia, palliative care, and NHs; multiple logistical pieces required to implement the intervention; inflexibility of palliative care providers to meet NH needs; and inability to assess residents in person. Facilitators included convenient, user-friendly and readily available telehealth equipment, and NH staff presence during visits. Outcomes most relevant to the key informants were increased goals of care conversations, improved symptom management and quality of life, and decreased health care utilization. Suggested adaptations included increased family engagement in the logistics of the intervention and strong NH advocacy.ConclusionsIn this study, key informants provided feedback that barriers to implementing NH telehealth palliative care far outweighed the facilitators for uptake. Future work will focus on employing NH staff in user centered design to overcome barriers such as optimal timing for consults and/or scheduled consult days to fit NH workflow, assessing organizational readiness for implementing change, and identifying dementia-specific and palliative care education needs.

BackgroundHip fractures significantly impact older adults, leading to compromised mobility and various adverse outcomes. The importance of early post-surgery mobilization in regaining pre-fracture levels of mobility is recognized, but lacks standardized definitions and implementation strategies. This study aimed to assess the impact of early sitting position 24 h after hip-fracture surgery on functional mobility recovery after 30 days using data from the Spanish National Hip Fracture Registry (RNFC).MethodsProspective cohort study, including patients aged ≥ 75 years admitted for hip-fracture surgery between 2017 and 2020 at Sant Camil Residential Hospital. Data from the RNFC were analyzed, and linear regression models were developed to assess the association between early sitting after surgery (ESAS) and mobility recovery at 30 days after surgery.ResultsOf 486 identified patients, 321 were included, with an estimated ESAS prevalence of 38.32% (95% CI: 32.97–43.88). ESAS was significantly associated with improved mobility recovery at 30 days. Multivariate regression models consistently revealed ESAS as a modest independent predictor of better post-surgery mobility. Factors such as age, cognitive capacity, and general health also impacted mobility recovery.ConclusionThe ESAS effect, while modest, emerges as a significant predictor of hip mobility recovery among older patients with hip fractures 30 days after surgery. These findings underscore the potential of this low-risk, low-cost intervention in enhancing functional mobility recovery strategies and emphasize the need for further research to uncover its broader implications in post-operative care. Implementation of early sitting could be enhanced, as only a third of patients in our study underwent this simple intervention.

BackgroundWhen admitted to hospital for unplanned medical needs, the complexity of multiple conditions, including cognitive and mental health, might put older people at greater risk, affecting their survival. This study aimed to investigate the prevalence of cognitive impairment versus cognitive impairment with depression and their association with six-month mortality in older people after an unplanned hospital admission in Italy.MethodsIn Re.Po.SI. a multi-centre study performed in Italy, standardized web-based case report forms were used to collect data on socio-demographic factors, clinical parameters, diagnoses, treatment history and at discharge, clinical events during hospitalization, and outcome data was collected. A comprehensive geriatric assessment was conducted using Cumulative Illness Rating Scale (CIRS), Geriatric Depression Scale (GDS-4), Barthel Index, and Short Blessed Test (SBT). To explore the interrelationship between depression and cognitive impairment, a variable categorized the study population into four mutually exclusive groups. This variable assessed the association between its categories and six-month mortality in a Cox multivariate analysis.ResultsOne thousand nine hundred fifty six participants were included, with a median age of 80 years (IQR: 73–85). Those who died within six months were likely to be older (82 vs. 79 years), male (56.2% vs. 47.2%), had moderately reduced ability to perform daily activities (82.0 vs. 93.0), exhibited greater illness severity (CIRS-IS: 1.8 vs. 1.6), had more chronically prescribed medications (6.0 vs. 5.0), and had a worse SBT score (10.0 vs. 7.0). When stratified based on cognitive impairment and depression, one-third had neither condition (33.2%), 21.9% had depression, 20.7% had a cognitive impairment, and 24.3% had both conditions. Six-month mortality was higher among people with cognitive impairment only (33.2%) followed by those with both conditions (28.8%), and depression only (22.7%). The unadjusted semi-parametric survival analysis revealed that the hazard ratio (HR) for people with cognitive impairment only was 2.08, for those with both conditions HR was 1.75, and for people with depression only HR was 1.30.ConclusionWhile depression alone may contribute to mortality risk, cognitive impairment appears to play a more substantial role in increasing the risk of dying within 6 month from an acute hospitalization. Further research is needed to confirm these finding.

BackgroundDepression has a detrimental effect on an individual’s mental and musculoskeletal strength multiplying the risk of fall incidents. The current study aims to investigate the association between depression and falls in older adults using machine learning (ML) approach and identify its various predictors.MethodsData for the study was derived from the Longitudinal Ageing Study in India, (LASI) conducted in 2017–18 for people aged 45-years and above. The study was carried out on 44,066 individuals. Depression was measured using the CIDI-SF scale. Bivariate cross-tabulations were used to study the prevalence of falls. And its association with depression and other independent factors were assessed using the novel ML, the Conditional inference trees (CIT) method.ResultsAround 10.8 percent of older adults had fall incidents. CIT model predicted region to be a significant predisposing factor for an older adult to experience falls. Multimorbidity, depression, sleep problems, and gender were other prominent factors. The model predicted that, among depressed older adults, falls incidents were around 80 percent higher than non-depressed.ConclusionsAn association between falls and depression was observed. Depressive symptoms were associated with an increased risk of falls, even after controlling for other co-factors. The CIT method leveraged us to select the most important variables to predict falls with great precision. To prevent and manage falls among the expanding and diverse older-aged population, a multilevel and cross-sectoral approach is required. Mental health, especially depression, should be dealt with greater precautions. Public health enthusiasts should focus on the physical as well as mental health of the country's older adult population.Graphical

Identifying cognitive impairment at an early stage is important to enable preventive treatment and lifestyle changes. As gait deviations precede cognitive impairment, the aim of this study was to investigate if step parameters during different Timed Up and Go (TUG) conditions could discriminate between people with different cognitive ability. Participants (N = 304) were divided into the following groups: (1) controls, n = 50, mean age:73, 44% women; (2) Subjective cognitive Impairment (SCI), n = 71, mean age:67, 45% women; (3) Mild Cognitive Impairment (MCI), n = 126, mean age: 73, 42% women; and (4) dementia disorders, n = 57, mean age: 78, 51% women. Participants conducted TUG and two motor-cognitive TUG-conditions: TUG while naming animals (TUGdt-NA) and reciting months in reverse order (TUGdt-MB). Tests were video recorded for data extraction of valid spatiotemporal parameters: step length, step width, step duration, single step duration and double step duration. Step length was investigated with the step length/body height ratio (step length divided by body height). Logistic regression models (adjusted for age, sex and education) investigated associations between step parameters and dichotomous variables of groups adjacent in cognitive ability: dementia disorders vs. MCI, MCI vs. SCI, and SCI vs. controls. Results were presented as standardized odds ratios (sORs), with 95% confidence intervals (CI95) and p-values (significance level: p < 0.05). The areas under the Receiver Operating Characteristic curves were presented for the step parameters/conditions with the highest sORs and, where relevant, optimal cutoff values were calculated. Step length showed greatest overall ability to significantly discriminate between adjacent groups (sOR ≤ . 67, CI95: .45-.99, p = ≤ . 047) during all group comparisons/conditions except three. The highest sOR for step-length was obtained when discriminating between SCI vs controls during TUGdt-MB (sOR = .51, CI95:.29- .87, p = .014), whereby the area under the curve was calculated (c-statistics = .700). The optimal cut-off indicated a step length of less than 32.9% (CI95 = 22.1-43.0) of body height to identify SCI compared with controls. The results indicate that step length may be important to assess during TUG, for discrimination between groups with different cognitive ability; and that the presented cut-off has potential to aid early detection of cognitive impairment. NCT05893524 (retrospectively registered 08/06/23).