Matthew P. Buman

This chapter provides an overview of the landscape of digital physical activity interventions, encompassing their evolution from early web-based programmes to sophisticated mHealth applications, wearable devices, and gamified approaches. It examines the evidence for their effectiveness, noting the potential for small to moderate improvements in physical activity behaviours, while also acknowledging the nuances across different modalities and the need for more long-term studies. Furthermore, the chapter discusses key theoretical frameworks and models guiding the development and implementation of these interventions, emphasising user-centred design and the use of well-established behaviour change theories. It also introduces intervention types at the forefront of the field, such as just-in-time adaptive interventions (JITAIs), and research study designs required to develop them. Looking ahead, future directions for digital physical activity interventions include a critical focus on addressing the digital divide and ensuring equitable access and effectiveness, enhancing data security and privacy, navigating the complexities of commercialisation, and leveraging the potential of emerging technologies like artificial intelligence and real-time responsive interventions like JITAIs to create more personalised and impactful programmes. Continued research utilising innovative study designs will be crucial to optimise these digital tools for broad public health benefit.

Background: Hispanic youth are disproportionately impacted by obesity and subsequent type 2 diabetes (T2D) yet remain underrepresented in diabetes prevention research. Digital health interventions hold promise for increasing the accessibility to and engagement in disease prevention programming, particularly among high-risk populations. However, there is a significant gap in the literature regarding digital T2D prevention programs for adolescents or Hispanic youth. Objective: The objective of this article is to describe the protocol for examining the feasibility of a 12-week digital diabetes prevention program among Hispanic adolescents with obesity. Methods: Participants (N=40; aged 12-16 years) will be randomized (1:1) to a 12-week intervention group or a control group. Youth in the intervention group will receive access to an e-learning platform with 12 nutrition and wellness video content sessions, a Fitbit Charge 5, and daily SMS text messages grounded in the self-determination theory to promote physical activity. Youth in the standard control group will receive information on diet and physical activity guidelines and guidance on setting behavior change goals. The study findings will focus on the evaluation of feasibility criteria: (1) recruitment of 40 Hispanic adolescents aged 12 to 16 years; (2) retention of 80% of the participants for postassessments; (3) integrity of the study protocol, defined as 70% or higher completion of content sessions and Fitbit wear on 5 days per week or more with response to 80% of SMS text messages when prompted; (4) 10% or lower incidence of technical issues; and (5) 80% or higher satisfaction among participants. Results: This study was funded in August 2022 and intervention implementation is ongoing. To date, 35 participants have been enrolled. Study findings will be available before December 2026 and will focus on an evaluation of the a priori feasibility criteria on participant recruitment, data collection, integrity of the study protocol, technical issues, and satisfaction. Study findings will also focus on secondary outcomes. Conclusions: The feasibility and process evaluation data obtained from this study will provide novel insights on the use of digital T2D prevention strategies among Hispanic youth and families and will inform the development of future digital health interventions among high-risk pediatric populations.

Background : Accelerometers are widely used to assess physical behaviors including physical activity, sedentary behavior, and sleep. However, inconsistent reporting of accelerometer methods limits comparability and reproducibility of findings. As part of developing a reporting guideline, we aimed to establish international consensus on key information that should be reported when describing accelerometry for assessing physical behaviors in health research. Methods : A modified eDelphi survey was conducted over two online rounds (November 2024 to April 2025). Participants included researchers, clinicians, journal editors, and other experts in accelerometry from 25 countries. Participants rated the importance of proposed reporting items using a 9-point Likert scale (1–3 = not important , 4–6 = important but not critical , and 7–9 = critical ) and were invited to provide further comments and suggestions for additional items. In Round 2, participants reviewed anonymized group feedback before rerating items. Consensus was defined as ≥70% rating an item as “critical.” Results : Of 122 registered participants, 111 (91%) completed Round 1, and 93 (76%) completed both rounds. Of 50 items rated, 39 (78%) met the predefined consensus threshold for inclusion. Most included items reflected core methodological details, such as device type, placement, and data processing procedures. Eleven items (21%) did not reach consensus, reflecting differing views on their importance (important vs. critical) and feasibility of reporting them within current practice. Conclusion : The 39 items identified as critically important provide a preliminary core set of reporting requirements to improve transparency, reproducibility, and comparability across studies. These will be refined and finalized through a consensus meeting and piloting.

Wearable sensors have found numerous applications in health and wellness promotion and have achieved great success leveraging advancements in deep learning. However, the development of robust continues to be hindered by issues related to sensor noise, inconsistent sampling rates, and individual differences. Topological data analysis (TDA) has emerged as a viable solution to extract robust features from such time-series data by converting them into persistence images (PIs), which capture intrinsic characteristics and demonstrate resilience to noise and signal variations. However, the computational costs of TDA pose significant challenges for small devices with limited resources. To more efficiently incorporate topological features, we utilize knowledge distillation (KD), which is a promising way to generate a smaller model using larger models. Multiple teachers can be adopted to enrich features in KD. However, this approach has presented two key challenges: 1) differences in feature dimensions from multimodal data and 2) conflicting knowledge provided by the different teachers, both of which can degrade the student model’s performance. To address these issues, we propose a novel KD framework called multimodal global latent workspace-based KD (mGLW-KD) that is motivated by global workspace theory (GTW) from cognitive neuroscience. GWT models how the brain integrates and distributes relevant information across different neural modules through a shared workspace, and it includes attentional control and working memory to prioritize and retain key information. Inspired by this theory, mGLW-KD incorporates a working memory module to unify diverse knowledge from multiple teacher models into a shared latent workspace, facilitating efficient knowledge transfer to the student model. By integrating topological insights with cognitive principles, mGLW-KD addresses the challenges posed by wearable sensor data and enables the student model to achieve superior performance using only time-series input during inference.

A common challenge for decision makers is selecting actions whose rewards are unknown and evolve over time based on prior policies. For instance, repeated use may reduce an action's effectiveness (habituation), while inactivity may restore it (recovery). These nonstationarities are captured by the Reducing or Gaining Unknown Efficacy (ROGUE) bandit framework, which models real-world settings such as behavioral health interventions. While existing algorithms can compute sublinear regret policies to optimize these settings, they may not provide sufficient exploration due to overemphasis on exploitation, limiting the ability to estimate population-level effects. This is a challenge of particular interest in micro-randomized trials (MRTs) that aid researchers in developing just-in-time adaptive interventions that have population-level effects while still providing personalized recommendations to individuals. In this paper, we first develop ROGUE-TS, a Thompson Sampling algorithm tailored to the ROGUE framework, and provide theoretical guarantees of sublinear regret. We then introduce a probability clipping procedure to balance personalization and population-level learning, with quantified trade-off that balances regret and minimum exploration probability. Validation on two MRT datasets concerning physical activity promotion and bipolar disorder treatment shows that our methods both achieve lower regret than existing approaches and maintain high statistical power through the clipping procedure without significantly increasing regret. This enables reliable detection of treatment effects while accounting for individual behavioral dynamics. For researchers designing MRTs, our framework offers practical guidance on balancing personalization with statistical validity.

With the increased demand for wearable sensors, image representations—such as persistence images and Gramian angular fields—transformed from time-series data have been investigated to address challenges in wearables arising from physiological variations, sensor noise, and limitations in capturing contextual information. To preserve the lightweight structural design of models, knowledge distillation (KD) has also been employed alongside image representations during training to distill smaller and more efficient models. Although image representations play a key role in providing richer and more informative features in training a model, their effectiveness within the KD framework has not been thoroughly explored. In this paper, we focus on image representation-driven KD to investigate whether these representations can provide useful knowledge leading to improved time-series interpretation in activity classification tasks. We explore the benefits of integrating image representations into KD, and we analyze the interplay between representation richness and model compactness with different combinations of teacher and student networks. We also introduce diverse KD strategies to utilize image representations, and we demonstrate the strategies with various perspectives, such as analysis of noises, generalizability, and compatibility, across datasets of varying scales to obtain comprehensive and insightful observations. These offer valuable insights for designing efficient and high-performance wearable sensor-based systems.

Postprandial hyperglycemia, marked by the blood glucose level exceeding the normal range after consuming a meal, is a critical indicator of progression toward type 2 diabetes in people with prediabetes and in healthy individuals. A key metric for understanding blood glucose dynamics after eating is the postprandial Area Under the Curve (AUC). Predicting postprandial AUC in advance based on a person’s lifestyle factors, such as diet and physical activity level, and explaining the factors that affect postprandial blood glucose could allow an individual to adjust their behavioral choices accordingly to maintain normal glucose levels. In this work, we develop an explainable machine learning solution, GlucoLens, that takes sensor-driven inputs and utilizes advanced data processing, large language models, and trainable machine learning models to estimate postprandial AUC and predict hyperglycemia from diet, physical activity, and recent glucose patterns. We use data obtained using wearables in a five-week clinical trial of 10 adults who worked full-time to develop and evaluate the proposed computational model that integrates wearable sensing, multimodal data, and machine learning. Our machine learning model takes multimodal data from wearable activity and glucose monitoring sensors, along with food and work logs, and provides an interpretable prediction of the postprandial glucose patterns. GlucoLens achieves a normalized root mean squared error (NRMSE) of 0.123 in its best configuration. On average, the proposed technology provides a 16% better predictive performance compared to the comparison models. Additionally, our technique predicts hyperglycemia with an accuracy of 79% and an F1 score of 0.749 and recommends different treatment options to help avoid hyperglycemia through diverse counterfactual explanations. With systematic experiments and discussion supported by established prior research, we show that our method is generalizable and consistent with clinical understanding.

Background: Using the multiphase optimization strategy (MOST) framework, we aimed to identify an optimized mHealth-delivered intervention for reducing recreational sedentary screen time (rSST) by at least 60 min/day among adults. Methods: Eligible participants were 23–64 years old and self-reported elevated rSST (> 3 h/day). Following a 7-day baseline, participants received a core mHealth application (self-monitoring and 50% reduction target and educational materials) and were randomly assigned to three additional components set to on/off in a full-factorial (2³) experiment: LOCKOUT: rSST electronically restricted; TEXT: rSST reduction prompts; and EARN: rSST through physical activity. rSST was assessed at baseline and 16 weeks via an integrated measure that included objectively assessed sedentary time (activPAL accelerometer) and screen time (TV Wifi plugs and tablet usage). We used a linear mixed effect model to evaluate the change in rSST for the three intervention components and their interactions. Results: A total of 82% of the randomized participants (N = 110) were female, with a mean ± SD age of 41 ± 11.7 y and a BMI of 29.7 ± 7.8 kg/m2, and their mean (95% CI) rSST was 184.7 (172.8, 196.5) min/day at baseline. The expected difference (baseline vs. 16 weeks) in rSST was greatest for the intervention versions with the core plus EARN on with an average reduction of -118.1 (-163.0, -73.1) min/day and for core plus LOCKOUT, TEXT, & EARN on (-125.7 [-172.0, -79.3] min/day). Conclusions: We identified several promising intervention versions that exceeded our optimization objective. This study provides important evidence on efficacious multicomponent interventions that should be moved forward to the evaluation phase of the MOST framework to test the effect of rSST reductions on health outcomes. Trial registration: (clinicaltrials.gov NCT04464993)