Abstract:Sensor devices have been increasingly used in engineering and health studies recently, and the captured multi-dimensional activity and vital sign signals can be studied in association with health outcomes to inform public health. The common approach is the scalar-on-function regression model, in which health outcomes are the scalar responses while high-dimensional sensor signals are the functional covariates, but how to effectively interpret results becomes difficult. In this study, we propose a new Functional Adaptive Double-Sparsity (FadDoS) estimator based on functional regularization of sparse group lasso with multiple functional predictors, which can achieve global sparsity via functional variable selection and local sparsity via zero-subinterval identification within coefficient functions. We prove that the FadDoS estimator converges at a bounded rate and satisfies the oracle property under mild conditions. Extensive simulation studies confirm the theoretical properties and exhibit excellent performances compared to existing approaches. Application to a Kinect sensor study that utilized an advanced motion sensing device tracking human multiple joint movements and conducted among community-dwelling elderly demonstrates how the FadDoS estimator can effectively characterize the detailed association between joint movements and physical health assessments. The proposed method is not only effective in Kinect sensor analysis but also applicable to broader fields, where multi-dimensional sensor signals are collected simultaneously, to expand the use of sensor devices in health studies and facilitate sensor data analysis.

What problem does this paper attempt to address?

This paper attempts to address the issue of effectively interpreting the relationship between high-dimensional functional covariates and scalar responses in multi-dimensional sensor data. Specifically, the authors focus on handling multiple functional covariates in the Functional Linear Regression Model and propose a new method called the Functional Adaptive Double-Sparsity estimator (FadDoS). ### Background and Problem With the widespread application of sensor devices in engineering and health research, the captured multi-dimensional activity and vital sign signals can be associated with health outcomes to guide public health. A commonly used method is the Scalar-on-Function Regression Model, where health outcomes are scalar responses, and high-dimensional sensor signals are functional covariates. However, effectively interpreting these results becomes very challenging. ### Research Objective The main objective of this paper is to propose a new Functional Adaptive Double-Sparsity estimator (FadDoS), which is based on Sparse Group Lasso with Multiple Functional Predictors. The FadDoS estimator achieves global sparsity through functional variable selection and local sparsity through the identification of zero sub-intervals within the coefficient functions. ### Methods 1. **Global Sparsity and Local Sparsity**: - Global Sparsity: Determine which functional covariates do not contribute to the response over the entire time domain through functional variable selection. - Local Sparsity: Further refine which sub-intervals within the coefficient functions do not affect the response by identifying zero sub-intervals. 2. **Functional Sparse Group Lasso Regularization**: - A new regularization method is proposed, combining Functional Sparse Group Lasso (fSGL) with adaptive weights to achieve flexible regularization. 3. **Algorithm Implementation**: - Use B-spline basis functions to represent the coefficient functions and solve the optimization problem through the Alternating Direction Method of Multipliers (ADMM). ### Theoretical Results - It is proven that the FadDoS estimator converges at a bounded rate under mild conditions and satisfies the oracle property. - Extensive simulation studies validate the theoretical properties and demonstrate superior performance compared to existing methods. ### Application - The FadDoS estimator is applied to a Kinect sensor study, which uses advanced motion-sensing devices to track multiple joint movements of the elderly, demonstrating how the FadDoS estimator effectively describes the detailed association between joint movements and physical health. - This method is not only applicable to Kinect sensor analysis but also to other fields that simultaneously collect multi-dimensional sensor signals, expanding the application of sensor devices in health research and promoting sensor data analysis. ### Conclusion The FadDoS estimator proposed in this paper significantly enhances the simplicity and interpretability of models when handling high-dimensional sensor data, addressing the issue of effectively interpreting complex data that existing methods struggle with.

Functional Adaptive Double-Sparsity Estimator for Functional Linear Regression Model with Multiple Functional Covariates

Linear Dynamic Sparse Modelling for Functional MR Imaging.

Smooth and Locally Sparse Estimation for Multiple-Output Functional Linear Regression

Covariate Dependent Sparse Functional Data Analysis

Functional Linear Regression with Mixed Predictors

Functional Regression with Mode-Sparsity Constraint

Factor-guided estimation of large covariance matrix function with conditional functional sparsity

Functional Data Analysis for Sparse Longitudinal Data

Fast covariance estimation for multivariate sparse functional data

Robust Shrinkage Estimation and Selection for Functional Multiple Linear Model Through LAD Loss

Regression Analysis of Asynchronous Longitudinal Functional and Scalar Data

Functional Linear Regression for Partially Observed Functional Data

Fast Covariance Estimation for Sparse Functional Data

Adaptive Weight and Wasserstein Distance Constrained Low-Rank Sparse Representation Method for Functional Connectivity Network Estimation

Variable selection in nonlinear function-on-scalar regression

Bayesian Function-on-Scalars Regression for High-Dimensional Data

Adaptive Functional Thresholding for Sparse Covariance Function Estimation in High Dimensions

Partially Functional Linear Regression in High Dimensions

Functional L-Optimality Subsampling for Massive Data

Variable Selection for Multiple Function-on-Function Linear Regression

High dimensional test for functional covariates