A novel basis expansion approach is proposed to fit a variable‐domain functional regression model, which assumes the functional nature of the covariate (daily steps) and incorporates the sample data into the model without alignment to enrich the interpretability of the model.
Abstract
ABSTRACT Physical activity plays an important role in the well‐being of patients with Chronic Obstructive Pulmonary Disease (COPD). The telEPOC program, developed at the Galdakao‐Usansolo University Hospital (Vizcaya, Spain), provided a telemonitoring data set resulting in a variable domain functional data set. In this sense, the aim is to study the impact of physical activity (measured as daily steps collected in a time interval delimited by the patient's participation in the experiment) on the rate of hospitalizations. To address this challenge, a novel basis expansion approach is proposed to fit a variable‐domain functional regression model, which assumes the functional nature of the covariate (daily steps) and incorporates the sample data into the model without alignment. This will enrich the interpretability of the model by incorporating the information provided by the domain through a two‐dimensional regression coefficient. In addition, a penalized estimation of the model is proposed in terms of a 2D P‐spline penalty, which allows considering different degree of smoothness for each dimension of the functional regression coefficient (anisotropic penalty). The performance of our methodology is also tested and compared with state‐of‐the‐art methods by means of a simulation study. Software for our method can be found in the VDPO R package available in CRAN.
In large-scale epidemiological studies, it is often of interest to investigate joint relationships between longitudinal and time-to-event outcomes with exposures that are trajectories or functions. Our motivation study is the Objective Physical Activity and Cardiovascular Health (OPACH) Study, which collected accelerometry-measured physical activity in 6,489 older women. One of the scientific aims is to understand sedentary behavior accumulation patterns and its association with physical function (longitudinal) and mortality (time-to-event). We propose a novel approach that first converts raw accelerometry data into daily sitting bout accumulation profiles, which are treated as functional covariates, and then develop a functional joint model for longitudinal and time-to-event outcomes that incorporates a baseline functional covariate. The longitudinal process is modeled using functional data methods and linked to the survival process through functional principal component scores. Both sub-models include interpretable linear scalar-on-function regression coefficients to capture flexible dose-response associations between sitting accumulated across varying bout durations and health outcomes. Estimation is carried out via an efficient expectation-maximization (EM) algorithm with penalized spline approximations. Simulation studies demonstrate accurate parameter estimation and reliable model selection. Application to the OPACH data reveals flexible and interpretable dose-response relationships between sitting bout durations, physical function, and mortality.
Luo Xiao, Wenyi Wang, Yumeng Zhang et al.· 0 citations
Physical activity (PA) is a pivotal modifiable determinant of intrinsic capacity (IC), a core component of healthy aging.
This study aimed to examine the association between PA and the risk of IC impairment among middle-aged and older adults using nationally representative cohorts from China and the United States.
Data from CHARLS (2011–2015) and HRS (2010–2018) were used, including 781 and 2,601 participants, respectively. PA was quantified by metabolic equivalents (MET), and IC impairment was defined as dysfunction in one or more of five domains. Cox proportional hazards regression, restricted cubic splines (RCS), subgroup analysis and nomograms were used for statistical analysis.
PA was inversely associated with the risk of IC impairment in both cohorts. In CHARLS, low and moderate PA significantly reduced IC impairment risk [Hazard Ratio (HR) = 0.691, 0.687;
P
< 0.05), while high-intensity PA had no significant effect. In HRS, all PA levels were protective (HR = 0.720, 0.736, 0.739;
P
< 0.05). RCS showed non-linear dose-response relationships between PA and IC impairment risk. Subgroup analysis indicated the protective effect of PA was consistent in CHARLS, but modified by marital status in HRS (only significant in married people). The nomograms showed modest discriminative ability [Area Under the Curve (AUC) > 0.6].
The protective association between PA and IC impairment varied by cohort and PA intensity.
PA was associated with a reduced risk of IC impairment, supporting targeted PA-based interventions for healthy aging.
Not applicable.
Xiao-Xian Yang, Yan Zhuang, Hui Huang· Aging Clinical and Experimen...· 0 citations
Physical activity levels are reduced in people with chronic obstructive pulmonary disease (COPD). The factors underpinning this, and its impact on individuals’ health, are complex. Our aim was to investigate whether within-person changes in physical activity are associated with subsequent health outcomes, encompassing six-minute walk distance, quadriceps strength, lung function, perceived fatigue (FACIT-F), and symptom burden (CAT).
In this multicentre cohort study, 565 individuals with COPD (37% female; age 68±8 years; FEV
1
54±21%pred) were followed over two years, with study visits every six months. Physical activity was assessed using wearable devices (Axivity AX6 or DynaPort MoveMonitor MM+), capturing daily step count and movement intensity during walking. Linear mixed-effects models assessed the longitudinal changes in physical activity and health outcomes, and the associations of within-person changes in physical activity and subsequent health outcomes.
Participants showed a significant annual decline of 229 (95%CI 132–326) steps/day. Correcting for baseline PA, each 1000-step annual decline was associated with a 5.13 m (95%CI 3.20–7.06) lower six-minute walk distance, 0.37%pred (95%CI 0.08–0.66) lower FEV
1
, and 0.47-point (95%CI 0.20–0.75) lower FACIT-F at the end of that year. Similar associations were observed for movement intensity as exposure. Only movement intensity was associated with functional quadriceps strength, with each 0.1 m·s
−2
decline resulting in a 0.18 s (95%CI 0.35–0.01) slower five-times sit-to-stand performance. No associations were found with the CAT.
Within-person decline in physical activity is associated with lower lung function, reduced exercise capacity, increased fatigue symptoms and lower functional quadriceps strength.
Paulien Mellaerts, H. Demeyer, C. Becker et al.· ERJ Open Research· 0 citations
OBJECTIVES
This study applied a robust predictive machine learning framework to a comprehensive set of individual background data to detect variables that are predictive of extremely low and high amounts of moderate to vigorous physical activity (MVPA). The main aim was to detect predictive variables, and as a secondary aim, the model accuracy was also assessed. Additionally, the analysis framework is presented for further use in physical activity research.
METHODS
Comprehensive data on health, physical functioning, education, work, and the living environment, represented by 130 variables, from 1536 participants aged from 20 to 69 years were used for analysis. The 2.5th and 97.5th quantiles of MVPA were chosen to represent the extreme physical behaviors in the population. Random forest classifiers were trained separately to predict the low and high MVPA groups. Results were confirmed and predictive variables detected with permutation tests and the Wilcoxon signed-rank test, Bonferroni-corrected for P < .05.
RESULTS
Nine significant predictors were found for the extremely low MVPA and 6 for the extremely high MVPA. These variables were related to the individual's mobility function, living environment, and occupation. The areas under the receiver operating characteristic curve were 0.73 and 0.59 for the participants with extremely low and high amounts of MVPA, respectively.
CONCLUSIONS
Several significant predictive variables for extremely low and high MVPA were discovered that may help characterize factors associated with PA behavior at the population level. However, it is not possible to accurately identify physically highly inactive or highly active individuals only from comprehensive background data.
S. Jauhiainen, Visa Nykänen, Sami Äyrämö et al.· Journal of Physical Activity...· 0 citations
Chronic obstructive pulmonary disease (COPD) remains a major cause of respiratory morbidity and healthcare utilization, creating challenges for healthcare planning, resource allocation, and early intervention. This study aimed to develop a forecasting framework for quarterly age- adjusted COPD outpatient visit rates across 762 regions in Pennsylvania from 2019 to 2023, using multi-source data including PHC4 outpatient records, satellite-derived environmental pollutant variables, and meteorological variables. To compare models that capture nonlinear relationships with those that explicitly model temporal dependencies, several machine learning models and a deep learning long short-term memory (LSTM) model, designed to learn sequential patterns and lagged temporal effects, were evaluated. The seasonal naïve baseline achieved R2=0.570, classical machine learning models achieved R2=0.58–0.62, and the LSTM model achieved R2=0.705 with lower prediction error. Lagged COPD activity was the strongest predictor, while environmental, meteorological, and geographic variables provided additional predictive information. These findings highlight the value of integrating multi-source environmental data with sequence-based modeling for forecasting regional COPD activity and suggest that environmental and meteorological variables can provide additional predictive information beyond historical COPD activity alone.
Physical activity (PA) is a core component of cancer survivorship care, yet adherence to PA guidelines remains low. Correlates of PA adherence reflect a complex interplay of multiple factors, many aligning with domains addressed in physical therapy (PT) but not integrated within a PT-informed analytic framework. Machine learning (ML) approaches may support modeling of complex relationships by accommodating nonlinear interactions. This study aimed to develop and compare ML models to classify PA adherence among adults with cancer using National Health and Nutrition Examination Survey (NHANES) data and a PT-informed feature selection framework.
This retrospective cross-sectional study included 1169 adults with a history of cancer from NHANES 2021-2023. From ~400 variables, PT-informed screening, redundancy reduction, and feature engineering yielded 45 features. Logistic regression, random forest, gradient boosting, XGBoost, and support vector machine (SVM) models were developed and compared. Discrimination was assessed using the area under the receiver operating characteristic curve (AUC). Permutation feature importance was applied to the best-performing model with a reduced-feature sensitivity analysis.
The SVM demonstrated the strongest performance (AUC = 0.749). Influential variables included self-rated health, sedentary time, educational attainment, difficulty walking or climbing steps, body mass index, gender, number of rooms in the home, systolic blood pressure, income-to-poverty ratio, and anxiety. A reduced SVM using the top 10 variables showed comparable discrimination (AUC = 0.757).
ML models demonstrated moderate discrimination in classifying PA adherence among adults with cancer. Parsimonious models using clinically accessible variables may support rehabilitation workflows by prompting targeted screening and timely intervention.
Qing Zhang, S. Gohel· Rehabilitation Oncology· 0 citations
Related blog posts
MIT News · Artificial Intelligence· news.mit.eduAug 17, 2026
A USAF cadet and a Lincoln Laboratory researcher found AI chatbots can help nontechnical service members produce viable software applications for their unique problems.
