MoveMentor—examining the effectiveness of a machine learning and app-based digital assistant to increase physical activity in adults: protocol for a randomised controlled trial

Increasing physical activity using an just-in-time adaptive digital assistant supported by machine learning: A novel approach for hyper-personalised mHealth interventions

Walkability is positively associated with physical activity in adults. Walkability is more consistently associated with walking for transportation than recreational walking. The purpose of this study is to examine how the association between walkable neighbourhoods and physical activity varies by age and type of physical activity using a new Canadian walkability database. The 2016 Canadian Active Living Environments (Can-ALE) database was attached to two cross-sectional health surveys: the Canadian Health Measures Survey (CHMS; 2009 to 2015) and the Canadian Community Health Survey (CCHS; 2015 to 2016). Physical activity was measured in the CHMS using the Actical accelerometer (n = 10,987; ages 3 to 79). Unorganized physical activity outside of school among children aged 3 to 11 was reported by parents in the CHMS (n = 4,030), and physical activity data by type (recreational, transportation-based, school-based, and household and occupational) was self-reported by respondents in the CCHS (n = 105,876; ages 12 and older). Walkability was positively associated with accelerometer-measured moderate-to-vigorous physical activity in youth (p < 0.05), younger adults (p < 0.0001) and older adults (p < 0.05), while walkability was negatively associated with light physical activity in youth (ages 12 to 17) and older adults (ages 60 to 79) (p < 0.05). Walkability was positively associated with self-reported transportation-based physical activity in youth (p < 0.001) and adults of all ages (p < 0.0001). Walkability was negatively associated with parent-reported unorganized physical activity of children aged 5 to 11, and children living in the most walkable neighbourhoods accumulated 10 minutes of physical activity less-on average-than those living in the least walkable neighbourhoods. The results of this study are consistent with previous studies indicating that walkability is more strongly associated with physical activity in adults than in children and that walkability is associated with transportation-based physical activity. Walkability is one of many built environment factors that may influence physical activity. More research is needed to identify and understand the built environment factors associated with physical activity in children and with recreational or leisure-time physical activity.

Measuring Physical Activity in Younger and Older Adults with Sickle Cell Disease Using Accelerometers

BackgroundInsufficient physical activity in the adult population is a global pandemic. Fun for Wellness (FFW) is a self-efficacy theory- and Web-based behavioral intervention developed to promote growth in well-being and physical activity by providing capability-enhancing opportunities to participants.ObjectiveThis study aimed to evaluate the effectiveness of FFW to increase physical activity in adults with obesity in the United States in a relatively uncontrolled setting.MethodsThis was a large-scale, prospective, double-blind, parallel-group randomized controlled trial. Participants were recruited through an online panel recruitment company. Adults with overweight were also eligible to participate, consistent with many physical activity–promoting interventions for adults with obesity. Also consistent with much of the relevant literature the intended population as simply adults with obesity. Eligible participants were randomly assigned to the intervention (ie, FFW) or the usual care (ie, UC) group via software code that was written to accomplish equal allocations to the FFW and UC groups. Data collection was Web based, fully automated, and occurred at three time points: baseline, 30 days after baseline (T2), and 60 days after baseline (T3). Participants (N=461) who were assigned to the FFW group (nFFW=219) were provided with 30 days of 24-hour access to the Web-based intervention. A path model was fit to the data consistent with the FFW conceptual model for the promotion of physical activity.ResultsThere was evidence for a positive direct effect of FFW on transport-related physical activity self-efficacy (beta=.22, P=.02; d=0.23), domestic-related physical activity self-efficacy (beta=.22, P=.03; d=0.22), and self-efficacy to regulate physical activity (beta=.16, P=.01; d=0.25) at T2. Furthermore, there was evidence for a positive indirect effect of FFW on physical activity at T3 through self-efficacy to regulate physical activity at T2 (beta=.42, 95% CI 0.06 to 1.14). Finally, there was evidence for a null direct effect of FFW on physical activity (beta=1.04, P=.47; d=0.07) at T3.ConclusionsThis study provides some initial evidence for both the effectiveness (eg, a positive indirect effect of FFW on physical activity through self-efficacy to regulate physical activity) and the ineffectiveness (eg, a null direct effect of FFW on physical activity) of the FFW Web-based behavioral intervention to increase physical activity in adults with obesity in the United States. More broadly, FFW is a scalable Web-based behavioral intervention that may effectively, although indirectly, promote physical activity in adults with obesity and therefore may be useful in responding to the global pandemic of insufficient physical activity in this at-risk population. Self-efficacy to regulate physical activity appears to be a mechanism by which FFW may indirectly promote physical activity in adults with obesity.Trial RegistrationClinicalTrials.gov NCT03194854; https://clinicaltrials.gov/ct2/show/NCT03194854.

Getting AI Right: Introductory Notes on AI &amp; Society

Purpose The purpose was to explore interrelations between factors related to engagement in physical activity in inactive adults with knee pain. Method Inactive adults with knee pain (n = 35) participated in six focus groups designed to inquire about barriers and facilitators related with engagement in physical activity. Directed content analysis and inductive thematic analysis were used to identify factors related to physical activity and associated interrelations respectively. As an exploratory analysis, sex differences in barriers and facilitators to physical activity were assessed. Results In this cohort (age = 60.9 ± 8.6 years; 22 females), self-reported physical activity was 26.3 ± 46.8 min/week. Factors related to physical activity were grouped into domains of physical status, psychological status, environment, knowledge, and resources. It was seen that the interrelations between a person and their environment, as well as, between impairments and everyday responsibilities influenced engagement in physical activity. Females were more likely to identify physical and psychological status, social expectations, and lack of knowledge as barriers. Males indicated a preference for using mobile technologies to overcome barriers. Conclusion Interplay of various barriers and facilitators is related to engagement in physical activity in inactive older adults with knee pain. Interventions to promote physical activity should address these interrelations and sex differences. Implications for rehabilitation Interrelations between individual factors related to engagement in physical activity and sex differences in these factors are present in inactive adults with knee pain. Interventions to improve physical activity should be implemented by addressing factors and interrelations between factors related to physical activity in inactive adults with knee pain. Interventions to address low levels of physical activity in adults with knee pain should take into account sex differences.

Classical machine learning algorithms typically operate on unimodal data and hence it can analyze and make predictions based on data from a single source (modality). Whereas multimodal machine learning algorithm, learns from information across multiple modalities, such as text, images, audio, and sensor data. The paper leverages the functionalities of multimodal machine learning (ML) application for generating text from images. The proposed work presents an innovative multimodal algorithm that automates the creation of news articles from geo-tagged images by leveraging cutting-edge developments in machine learning, image captioning, and advanced text generation technologies. Employing a multimodal approach that integrates machine learning and transformer algorithms, such as visual geometry group network16 (VGGNet16), convolutional neural network (CNN) and a long short-term memory (LSTM) based system, the algorithm initiates by extracting the location from exchangeable image file format (Exif) data from the image. The features are extracted from the image and corresponding news headline is generated. The headlines are used for generating a comprehensive article with contemporary large language model (LLM). Further, the algorithm generates the news article big-science large open-science open-access multilingual language model (BLOOM). The algorithm was tested on real time photographs as well as images from the internet. In both the cases the news articles generated were validated with ROUGE and BULE score. The proposed work is found to be successful attempt in journalism field.

The aim of this study was to examine the level of physical activity in adults with cerebral palsy (CP) and to analyse its relationship with physical activity as adolescents, pain, and gross motor function. A prospective cohort study was performed using data from the Swedish National CP Registry (CPUP) for all 129 individuals born in 1991–1993 living in Skåne and Blekinge who reported to CPUP at 14–16 years of age. Physical activity as adult was analysed relative to physical activity as adolescents, pain, and the Gross Motor Function Classification System (GMFCS). Seventy-one individuals at GMFCS I–V were followed up as adults and included in the analyses. Of these, 65% were physically active, but only 56% performed physical activity at least once a week. Their physical activity as adults differed relative to their physical activity as adolescents (p = 0.011) but not to pain or GMFCS. Being physically active as an adolescent doubled the probability of being active as an adult (OR 2.1; p = 0.054), indicating that physical activity in adults with CP is related to their physical activity as adolescents. Therefore, interventions to increase physical activity among adolescents with CP are likely also to improve physical activity in adulthood.

Electrochemical C−H oxidation reactions offer a sustainable route to functionalize hydrocarbons, yet identifying suitable substrates and optimizing synthesis remain challenging. Here, we report an integrated approach combining machine learning and large language models to streamline the exploration of electrochemical C−H oxidation reactions. Utilizing a batch rapid screening electrochemical platform, we evaluated a wide range of reactions, initially classifying substrates by their reactivity, while LLMs text‐mined literature data to augment the training set. The resulting ML models for reactivity prediction achieved high accuracy (&gt;90 %) and enabled virtual screening of a large set of commercially available molecules. To optimize reaction conditions for selected substrates, LLMs were prompted to generate code that iteratively improved yields. This human‐AI collaboration proved effective, efficiently identifying high‐yield conditions for 8 drug‐like substances or intermediates. Notably, we benchmarked the accuracy and reliability of 12 different LLMs–including LLaMA series, Claude series, OpenAI o1, and GPT‐4‐on code generation and function calling related to ML based on natural language prompts given by chemists to showcase potentials for accelerating research across four diverse tasks. In addition, we collected an experimental benchmark dataset comprising 1071 reaction conditions and yields for electrochemical C−H oxidation reactions.

The role of physical activity in diabetes is critical, influencing this disease's development, man-agement, and overall outcomes. In China, 22.3% of adults do not meet the minimum level of physical activity recommended by the World Health Organization. Therefore, it is imperative to identify the factors that contributing to lack of physical activity must be identified. To investigate the relationship among delay discounting, delay aversion, glycated hemoglobin (HbA1c), and various levels of physical activity in Chinese adults diagnosed with type 2 diabetes mellitus (T2DM). In 2023, 400 adults with T2DM were recruited from the People's Hospital of Linxia Hui Autonomous Prefecture of Gansu Province. A face-to-face questionnaire was used to gather demographic data and details on physical activity, delay discounting, and delay aversion. In addition, HbA1c levels were measured in all 400 participants. The primary independent variables considered were delay discounting and delay aversion. The outcome variables included HbA1c levels and different intensity levels of physical activity, including walking, moderate physical activity, and vigorous physical activity. Multiple linear regression models were utilized to assess the relationship between delay discounting, delay aversion, and HbA1c levels, along with the intensity of different physical activity measured in met-hours per week. After controlling for the sample characteristics, delay discounting was negatively associated with moderate physical activity (β = -2.386, 95%CI: -4.370 to -0.401). Meanwhile, delay aversion was negatively associated with the level of moderate physical activity (β = -3.527, 95% CI: -5.578 to -1.476) in the multiple linear regression model, with statistically significant differences. Elevated delay discounting and increased delay aversion correlated with reduced levels of moderate physical activity. Result suggests that delay discounting and aversion may influence engagement in moderate physical activity. This study recommends that health administration and government consider delay discounting and delay aversion when formulating behavioral intervention strategies and treatment guidelines involving physical activity for patients with T2DM, which may increase participation in physical activity. This study contributes a novel perspective to the research on physical activity in adults with T2DM by examining the significance of future health considerations and the role of emotional responses to delays.

BackgroundVigorous intermittent lifestyle physical activity (VILPA; short bursts of vigorous-intensity activities in a person’s daily life) could be an attractive and feasible option to increase physical activity (PA) in adults transitioning to retirement.Design and settingTwo-arm pilot randomised controlled trial (RCT) to test the feasibility of the intervention and the plausibility of the intervention to increase PA in adults transitioning to retirement in Perth, Western Australia.ParticipantsInsufficiently physically active adults transitioning to retirement.InterventionTwelve-week theory-based and evidence-informed VILPA intervention designed to increase PA in adults transitioning to retirement.Objectives and measurementsThe feasibility of the pilot was determined by the projected sample size with actual sample size, drop-out rates and reporting rates. The feasibility, acceptability and appropriateness of the intervention were assessed using validated questionnaires. The intervention’s plausibility to increase PA was assessed by accelerometer-measured PA, functional fitness test and general health questionnaire.ResultsEighty individuals expressed interest in participating in the trial; 42 (feasibility of recruitment = 52.5%) were recruited and 34 completed the trial (retention = 80%). The preliminary data indicated increases in both total PA and VILPA, with positive impacts in self-reported general health and functional fitness. Participants found the intervention acceptable and intended to continue participation in VILPA and accumulate PA after the intervention.ConclusionsThe VILPA intervention appears to be feasible for promoting PA in ageing adults. The findings of this pilot RCT also support a larger trial to seek the effectiveness of VILPA in improving health outcomes in ageing adults.

Pushing forward: Understanding physical activity in adults with medical complexity.

Electrochemical C-H oxidation reactions offer a sustainable route to functionalize hydrocarbons, yet identifying suitable substrates and optimizing synthesis remain challenging. Here, we report an integrated approach combining machine learning and large language models to streamline the exploration of electrochemical C-H oxidation reactions. Utilizing a batch rapid screening electrochemical platform, we evaluated a wide range of reactions, initially classifying substrates by their reactivity, while LLMs text-mined literature data to augment the training set. The resulting ML models for reactivity prediction achieved high accuracy (>90 %) and enabled virtual screening of a large set of commercially available molecules. To optimize reaction conditions for selected substrates, LLMs were prompted to generate code that iteratively improved yields. This human-AI collaboration proved effective, efficiently identifying high-yield conditions for 8 drug-like substances or intermediates. Notably, we benchmarked the accuracy and reliability of 12 different LLMs-including LLaMA series, Claude series, OpenAI o1, and GPT-4-on code generation and function calling related to ML based on natural language prompts given by chemists to showcase potentials for accelerating research across four diverse tasks. In addition, we collected an experimental benchmark dataset comprising 1071 reaction conditions and yields for electrochemical C-H oxidation reactions.

Background and Aims Large language models (LLMs) have gained significant attention in the field of natural language processing (NLP), marking a shift from traditional techniques like Term Frequency-Inverse Document Frequency (TF-IDF). We developed a traditional NLP model to predict arteriovenous fistula (AVF) failure within next 30 days using clinical notes. The goal of this analysis was to investigate whether LLMs would outperform traditional NLP techniques, specifically in the context of predicting AVF failure within the next 30 days using clinical notes. Method We defined AVF failure as the change in status from active to permanently unusable status or temporarily unusable status. We used data from a large kidney care network from January 2021 to December 2021. Two models were created using LLMs and traditional TF-IDF technique. We used “distilbert-base-uncased”, a distilled version of BERT base model [1], and compared its performance with traditional TF-IDF-based NLP techniques. The dataset was randomly divided into 60% training, 20% validation and 20% test dataset. The test data, comprising of unseen patients’ data was used to evaluate the performance of the model. Both models were evaluated using metrics such as area under the receiver operating curve (AUROC), accuracy, sensitivity, and specificity. Results The incidence of 30 days AVF failure rate was 2.3% in the population. Both LLMs and traditional showed similar overall performance as summarized in Table 1. Notably, LLMs showed marginally better performance in certain evaluation metrics. Both models had same AUROC of 0.64 on test data. The accuracy and balanced accuracy for LLMs were 72.9% and 59.7%, respectively, compared to 70.9% and 59.6% for the traditional TF-IDF approach. In terms of specificity, LLMs scored 73.2%, slightly higher than the 71.2% observed for traditional NLP methods. However, LLMs had a lower sensitivity of 46.1% compared to 48% for traditional NLP. However, it is worth noting that training on LLMs took considerably longer than TF-IDF. Moreover, it also used higher computational resources such as utilization of graphics processing units (GPU) instances in cloud-based services, leading to higher cost. Conclusion In our study, we discovered that advanced LLMs perform comparably to traditional TF-IDF modeling techniques in predicting the failure of AVF. Both models demonstrated identical AUROC. While specificity was higher in LLMs compared to traditional NLP, sensitivity was higher in traditional NLP compared to LLMs. LLM was fine-tuned with a limited dataset, which could have influenced its performance to be similar to that of traditional NLP methods. This finding suggests that while LLMs may excel in certain scenarios, such as performing in-depth sentiment analysis of patient data for complex tasks, their effectiveness is highly dependent on the specific use case. It is crucial to weigh the benefits against the resources required for LLMs, as they can be significantly more resource-intensive and costly compared to traditional TF-IDF methods. This highlights the importance of a use-case-driven approach in selecting the appropriate NLP technique for healthcare applications.

BackgroundAcquired brain injury (ABI), often arising from stroke or trauma, is a common cause of long-term disability, physical inactivity and poor health outcomes globally. Individuals with ABI face many barriers to increasing physical activity, such as impaired mobility, access to services and knowledge regarding management of physical activity. Self-management programmes aim to build skills to enable an individual to manage their condition, including their physical activity levels, over a long period of time. Programme delivery modes can include traditional face-to-face methods, or remote delivery, such as via the Internet. However, it is unknown how effective these programmes are at specifically improving physical activity in community-dwelling adults with ABI, or how effective and acceptable remote delivery of self-management programmes is for this population.Methods/DesignWe will conduct a comprehensive search for articles indexed on MEDLINE, EMBASE, CINAHL, PsychINFO, AMED, Cochrane Central Register of Controlled Trials (CENTRAL), PEDro and Science Citation Index Expanded (SCI-EXPANDED) databases that assess the efficacy of a self-management intervention, which aims to enhance levels of physical activity in adults living in the community with ABI. Two independent reviewers will screen studies for eligibility, assess risk of bias, and extract relevant data. Where possible, a meta-analysis will be performed to calculate the overall effect size of self-management interventions on physical activity levels and on outcomes associated with physical activity. A comparison will also be made between face-to-face and remote delivery modes of self-management programmes, in order to examine efficacy and acceptability. A content analysis of self-management programmes will also be conducted to compare aspects of the intervention that are associated with more favourable outcomes.DiscussionThis systematic review aims to review the efficacy of self-management programmes aimed at increasing physical activity levels in adults living in the community with ABI, and the efficacy and acceptability of remote delivery of these programmes. If effective, remote delivery of self-management programmes may offer an alternative way to overcome barriers and empower individuals with ABI to increase their levels of physical activity, improving health and general wellbeing.Trial registrationOur protocol has been registered on PROSPERO 2013: CRD42013006748.