The widespread adoption of wearable sensors has the potential to provide massive and heterogeneous time series data, driving the use of Artificial Intelligence in human sensing applications. However, data collection remains limited due to stringent ethical regulations, privacy concerns, and other constraints, hindering progress in the field. Synthetic data generation, particularly through Generative Adversarial Networks and Diffusion Models, has emerged as a promising solution to mitigate both data scarcity and privacy issues. However, these models are often limited to narrow operational scenarios, such as short-term and unimodal signal patterns. To address this gap, we present a systematic evaluation of state-of-the-art generative models for time series data, explicitly assessing their performance in challenging scenarios such as stress and emotion recognition. Our study examines the extent to which these models can jointly handle multi-modality, capture long-range dependencies, and support conditional generation-core requirements for real-world wearable sensor data generation. To enable a fair and rigorous comparison, we also introduce an evaluation framework that evaluates both the intrinsic fidelity of the generated data and their utility in downstream predictive tasks. Our findings reveal critical limitations in the existing approaches, particularly in maintaining cross-modal consistency, preserving temporal coherence, and ensuring robust performance in train-on-synthetic, test-on-real, and data augmentation scenarios. Finally, we present our future research directions to enhance synthetic time series generation and improve the applicability of generative models in the wearable computing domain.

Wearable sensor technologies and deep learning are transforming healthcare management. Yet, most health sensing studies focus narrowly on physical chronic diseases. This overlooks the critical need for joint assessment of comorbid physical chronic diseases and depression, which is essential for collaborative chronic care. We conceptualize multi-disease assessment, including both physical diseases and depression, as a multi-task learning (MTL) problem, where each disease assessment is modeled as a task. This joint formulation leverages inter-disease relationships to improve accuracy, but it also introduces the challenge of double heterogeneity: chronic diseases differ in their manifestation (disease heterogeneity), and patients with the same disease show varied patterns (patient heterogeneity). To address these issues, we first adopt existing techniques and propose a base method. Given the limitations of the base method, we further propose an Advanced Double Heterogeneity-based Multi-Task Learning (ADH-MTL) method that improves the base method through three innovations: (1) group-level modeling to support new patient predictions, (2) a decomposition strategy to reduce model complexity, and (3) a Bayesian network that explicitly captures dependencies while balancing similarities and differences across model components. Empirical evaluations on real-world wearable sensor data demonstrate that ADH-MTL significantly outperforms existing baselines, and each of its innovations is shown to be effective. This study contributes to health information systems by offering a computational solution for integrated physical and mental healthcare and provides design principles for advancing collaborative chronic disease management across the pre-treatment, treatment, and post-treatment phases.

The analysis of wearable sensor data has enabled many successes in several applications. To represent the high-sampling rate time-series with sufficient detail, the use of topological data analysis (TDA) has been considered, and it is found that TDA can complement other time-series features. Nonetheless, due to the large time consumption and high computational resource requirements of extracting topological features through TDA, it is difficult to deploy topological knowledge in various applications. To tackle this problem, knowledge distillation (KD) can be adopted, which is a technique facilitating model compression and transfer learning to generate a smaller model by transferring knowledge from a larger network. By leveraging multiple teachers in KD, both time-series and topological features can be transferred, and finally, a superior student using only time-series data is distilled. On the other hand, mixup has been popularly used as a robust data augmentation technique to enhance model performance during training. Mixup and KD employ similar learning strategies. In KD, the student model learns from the smoothed distribution generated by the teacher model, while mixup creates smoothed labels by blending two labels. Hence, this common smoothness serves as the connecting link that establishes a connection between these two methods. In this paper, we analyze the role of mixup in KD with time-series as well as topological persistence, employing multiple teachers. We present a comprehensive analysis of various methods in KD and mixup on wearable sensor data.

Visual Question-Answering, a technology that generates textual responses from an image and natural language question, has progressed significantly. Notably, it can aid in tracking and inquiring about daily activities, crucial in healthcare monitoring, especially for elderly patients or those with memory disabilities. However, video poses privacy concerns and has a limited field of view. This paper presents Sensor2Text, a model proficient in tracking daily activities and engaging in conversations using wearable sensors. The approach outlined here tackles several challenges, including low information density in wearable sensor data, insufficiency of single wearable sensors in human activities recognition, and model's limited capacity for Question-Answering and interactive conversations. To resolve these obstacles, transfer learning and student-teacher networks are utilized to leverage knowledge from visual-language models. Additionally, an encoder-decoder neural network model is devised to jointly process language and sensor data for conversational purposes. Furthermore, Large Language Models are also utilized to enable interactive capabilities. The model showcases the ability to identify human activities and engage in Q\&A dialogues using various wearable sensor modalities. It performs comparably to or better than existing visual-language models in both captioning and conversational tasks. To our knowledge, this represents the first model capable of conversing about wearable sensor data, offering an innovative approach to daily activity tracking that addresses privacy and field-of-view limitations associated with current vision-based solutions.

Deep learning advancements have revolutionized scalable classification in many domains including computer vision. However, when it comes to wearable-based classification and domain adaptation, existing computer vision-based deep learning architectures and pretrained models trained on thousands of labeled images for months fall short. This is primarily because wearable sensor data necessitates sensor-specific preprocessing, architectural modification, and extensive data collection. To overcome these challenges, researchers have proposed encoding of wearable temporal sensor data in images using recurrent plots. In this paper, we present a novel modified-recurrent plot-based image representation that seamlessly integrates both temporal and frequency domain information. Our approach incorporates an efficient Fourier transform-based frequency domain angular difference estimation scheme in conjunction with the existing temporal recurrent plot image. Furthermore, we employ mixup image augmentation to enhance the representation. We evaluate the proposed method using accelerometer-based activity recognition data and a pretrained ResNet model, and demonstrate its superior performance compared to existing approaches.

This study aimed to develop a wearable sensor system, using machine-learning models, capable of accurately estimating peak ground reaction force (GRF) during ballet jumps in the field. Female dancers (n = 30) performed a series of bilateral and unilateral ballet jumps. Dancers wore six ActiGraph Link wearable sensors (100 Hz). Data were collected simultaneously from two AMTI force platforms and synchronised with the ActiGraph data. Due to sensor hardware malfunctions and synchronisation issues, a multistage approach to model development, using a reduced data set, was taken. Using data from the 14 dancers with complete multi-sensor synchronised data, the best single sensor was determined. Subsequently, the best single sensor model was refined and validated using all available data for that sensor (23 dancers). Root mean square error (RMSE) in body weight (BW) and correlation coefficients (r) were used to assess the GRF profile, and Bland–Altman plots were used to assess model peak GRF accuracy. The model based on sacrum data was the most accurate single sensor model (unilateral landings: RMSE = 0.24 BW, r = 0.95; bilateral landings: RMSE = 0.21 BW, r = 0.98) with the refined model still showing good accuracy (unilateral: RMSE = 0.42 BW, r = 0.80; bilateral: RMSE = 0.39 BW, r = 0.92). Machine-learning models applied to wearable sensor data can provide a field-based system for GRF estimation during ballet jumps.

CHICAGO – PhysIQ, a leader in applying artificial intelligence to wearable sensor data, today announced that it has received 510(k) clearance from the U.S. Food and Drug Administration (FDA) for their algorithm to continuously determine respiration rate in ambulatory patients. This clearance adds to their expanding portfolio of FDA-cleared cloud-based analytics, which also include QRS detection, heart rate, heart rate variability, atrial fibrillation detection, and their personalized physiology change detection analytic. The latest clearance advances physIQ's strategy to offer a deep portfolio of FDA-cleared analytics that can be applied to wearable sensor data. To enable this, physIQ's platform collects raw telemetry from the device and uploads it to the cloud where FDA-cleared analytics use the raw biosignals to produce vital signs. With this approach physIQ is able to provide vital sign analytics that benefit from the superior computing power of the cloud and fuel the higher-level analytics that further characterize dimensions of human physiology.

With the increasing availability of wearable devices, continuous monitoring of individuals' physiological and behavioral patterns has become significantly more accessible. Access to these continuous patterns about individuals' statuses offers an unprecedented opportunity for studying complex diseases and health conditions such as type 2 diabetes (T2D). T2D is a widely common chronic disease that its roots and progression patterns are not fully understood. Predicting the progression of T2D can inform timely and more effective interventions to prevent or manage the disease. In this study, we have used a dataset related to 63 patients with T2D that includes the data from two different types of wearable devices worn by the patients: continuous glucose monitoring (CGM) devices and activity trackers (ActiGraphs). Using this dataset, we created a model for predicting the levels of four major biomarkers related to T2D after a one-year period. We developed a wide and deep neural network and used the data from the demographic information, lab tests, and wearable sensors to create the model. The deep part of our method was developed based on the long short-term memory (LSTM) structure to process the time-series dataset collected by the wearables. In predicting the patterns of the four biomarkers, we have obtained a root mean square error of 1.67% for HBA1c, 6.22 mg/dl for HDL cholesterol, 10.46 mg/dl for LDL cholesterol, and 18.38 mg/dl for Triglyceride. Compared to existing models for studying T2D, our model offers a more comprehensive tool for combining a large variety of factors that contribute to the disease.