In battery pack assembly, cells are graded by measuring discharge capacity for consistent module performance. However, existing grading tests often consider the capacity at a single temperature, which does not reflect the different environmental conditions for battery packs in real-world use. In this work, we propose a robustness-aware capacity prediction approach (Ra-CPA) for efficient cell grading under multi-temperature conditions. The proposed method integrates a robust feature selection with features extracted from the charging, rest, and partial-discharge phases, while incorporating a noise-aware mechanism to mitigate sample-level variability, leading to accurate and robust capacity prediction. The proposed method thus can effectively improve grading robustness through capacity prediction across multiple temperatures. Furthermore, it can reduce testing time based on accurate capacity prediction with partial discharge testing, offering practical potential for quality control in battery pack manufacturing. The performance of the proposed methods is evaluated across three major lithium-ion chemistries.

Performance degradation modeling based on data fusion has been extensively explored for constructing health indicators (HIs) to support condition monitoring and performance assessment of industrial assets. However, most existing studies employ time-invariant weighting schemes for long-term degradation modeling and therefore rarely account for the time-varying characteristics of degradation features. In addition, many methods emphasize predictive performance and accuracy at the cost of interpretability, while their dependence on large numbers of fault samples restricts their practical engineering applicability. To address these challenges, this study proposes an interpretable optimization modeling approach with dynamic time-varying weighting for practical HI construction and degradation stage assessment. First, a time-varying weighting strategy is introduced to define a HI that dynamically captures the evolving contributions of individual spectral components throughout the degradation process. Then, an interpretable two-dimensional optimization model is developed to automatically determine the time-varying feature weights. Finally, a dynamic update framework for time-varying weights is designed to accommodate zero-fault-sample scenarios and further enhance the practical applicability of the proposed method during long-term performance assessment. Experimental results from two case studies demonstrate that the proposed dynamic weighting strategy can reflect the physical evolution of frequency components and strengthen fault-sensitive indications, thereby enabling interpretable early fault detection and diagnosis of industrial assets. Moreover, the developed HI exhibits clearer transition behaviors across degradation stages compared with HIs constructed using static weights and other state-of-the-art approaches.

Predicting the state-of-health (SOH) of electric vehicle (EV) batteries under real-world operating conditions remains challenging due to noisy measurements, irregular charging behaviors, and highly variable environments, which differ substantially from controlled laboratory settings. This study proposes a vision-powered, multi-modal generative paradigm for accurate EV battery SOH prediction, which reformulates numerical SOH signals into image-based visual representations and leverages pixel-level spatial correlations learned from clean data to reconstruct underlying degradation patterns from noisy EV measurements. The paradigm comprises four key stages: image synthesis, generative model construction, pixel mapping, and SOH prediction. The proposed generative framework integrates a U-Net-based visual channel with a numerical temporal channel, allowing local spatial degradation patterns to be effectively fused with multivariate temporal dynamics. Advanced attention mechanisms are further incorporated to selectively enhance degradation-related representations and to facilitate coherent cross-modal information fusion. The proposed method is evaluated on one laboratory dataset and three real-world EV datasets encompassing diverse battery chemistries, vehicle types, and operational conditions. Experimental results demonstrate that the proposed paradigm effectively mitigates data-quality limitations inherent in field measurements and achieves highly accurate SOH prediction, attaining a minimum root-mean-square error of 0.0013. Comprehensive benchmarking against 15 state-of-the-art methods further confirms its consistently superior generalization performance across heterogeneous datasets. Interpretability analyses further reveal that both visual and numerical modalities capture physics-consistent aging patterns. These findings highlight the potential of the proposed vision-powered paradigm to extract stable degradation signatures from noisy EV operational data and its applicability to broader health prognostics tasks.

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.

In machine fault diagnosis, conventional data-driven models trained by empirical risk minimization (ERM) often fail to generalize across domains with distinct data distributions caused by various machine operating conditions. One major reason is that ERM primarily focuses on informativeness of data labels and lacks sufficient attention on invariance of data features. To enable invariance on top of informativeness, a learning framework, learning invariant features via in-label swapping for generalizing out-of-distribution (Lifeisgood), is proposed in this study. Lifeisgood is inspired by a simple intuition that invariance can be assessed by checking changes in loss due to swapping certain entries of features with the same labels. Lifeisgood also enjoys a theoretical guarantee on improving testing domain performance under certain conditions based on a swapping 0-1 loss proposed in this work. To circumvent the training difficulties associated with the swapping 0-1 loss, a swapping cross-entropy loss is derived as a surrogate and theoretical justifications for such a relaxation are also provided. As a result, Lifeisgood can be employed conveniently to develop data-driven fault diagnosis models. In the experiments, Lifeisgood outperformed the majority of state-of-the-art methods in terms of average accuracy and exceeded the second-best by 25% in terms of the frequency of beating the generic ERM. The code is available at: https://github.com/mozhenling/doge-lifeisgood

Empirical risk minimization (ERM) is a celebrated induction principle for developing data-driven models. However, ERM has received both pros and cons for its capability on domain generalization (DG). To this end, this paper attempts to study the success and failure of ERM at supervised DG classification tasks, both theoretically and empirically, with causal perspectives. In the theoretical aspect, we first explore different properties of a causal metric termed information flow, followed with discussing relationships between the information flow and the mutual information in the proposed causal graph. Next, we analyze the roles of the transformed causal feature and the transformed spurious feature on modeling performances. It reveals that the interaction between the spurious influencer and the transformed causal feature is the key determining the failure or success of ERM on DG. In the empirical study, we first simulate various DG settings based on the MNIST, Fashion MNIST, and CIFAR10 datasets. Next, we verify developed theories by testing three different neural network configurations in designed experiments. In addition, experiments based on real-world datasets are conducted to further consolidate key points of the proposed theories. To extend application benefits of the theoretical discoveries, a new risk minimization framework with a novel feature intervention for regulating ERM is proposed. It achieves DG improvements over ERM on real-world datasets of image segmentation, image classification, and text classification.

Background Despite the eruption of digital care systems for older adults, their benefits and challenges in clinical practice are uncertain. Objective We aimed to explore physiotherapists’ perspectives on technology-based gait and balance assessment systems for older adults and provide exploration insights to inform potential design requirements and candidate metrics for future systems. Methods Qualitative research based on semi-structured interviews with 20 physiotherapists was conducted to examine physiotherapists’ expertise in gait and balance evaluations (e.g., clinical processes, tests, and metrics), constraints and obstacles during clinical practices, and perspectives on the essential attributes and functionalities of optimal technology-infused systems. Results Findings indicate the Berg Balance Scale and movements of the pelvis, hip, knee, and feet are crucial for assessment. In addressing the constraints of conventional clinical assessments, technology-driven platforms enable the ongoing surveillance of daily movements and the quantification of kinematic measures. Physiotherapists also emphasized the significance of technology-integrated systems, accentuating physiotherapy-led evaluations, safety protocols, and non-intrusiveness. Conclusion This study serves as an effort to bridge the gap between technological progressions and real-world implementations through the lens of physiotherapists.

Incorrect labels as well as the discrepancy between training and test domain data distributions can significantly affect the effectiveness of supervised data-driven models in machine fault diagnosis applications. Such a challenge can be characterized as the noisy label-domain generalization (NL-DG) problem. In this article, the extended invariant risk minimization (EIRM) is developed, which incorporates flat minima seeking to address the NL-DG challenge. The ability of handling NL-DG is realized by shifting the gradient penalty base from the dummy classifier to the entire model. EIRM is shown to be closely related to locating a flat minimum, which is crucial for label noise (LN) robustness and model generalization. Explorations on function smoothness and algorithm convergence are offered to understand EIRM from the theoretical aspect. An efficient implementation of EIRM is also developed to construct the fault diagnosis model. The EIRM-based fault diagnosis method is compared with strong benchmarks on multiple NL-DG tasks using actuator and gearbox fault datasets. Results indicate that the EIRM-based method on average is more effective than the benchmarks. The code is available at https://github.com/mozhenling/doge-eirm.

State of health (SOH) estimation of battery packs in electric vehicles (EVs) is essential for transportation electrification safety and reliability. The noise and complexity of EV battery pack data hinder the effectiveness of various data-driven SOH estimation methods using lab data. To address these challenges and achieve more effective data-driven EV battery pack SOH predictions, this study develops a comprehensive deep-learning-based SOH modeling framework for EV batteries. The framework begins with a two-stage mode decomposition (TSMD) method designed to effectively identify neat SOH degradation patterns better representing noisy field data. Next, an endogenous and exogenous multibranch network structure with a hierarchically fused attention mechanism (EEMB-HiFA) is developed for real-time prediction of EV battery pack SOH. Computational experiments leveraging datasets from seven EVs are conducted to validate the accuracy and adaptiveness of the proposed EEMB-HiFA. The results show that the EEMB-HiFA can achieve a 96.49% improvement in accuracy compared to strong benchmarks considered.

Functional ANOVA (FANOVA) is a widely used variance-based sensitivity analysis tool. However, studies on functional-output FANOVA remain relatively scarce, especially for black-box computer experiments, which often involve complex and nonlinear functional-output relationships with unknown data distribution. Conventional approaches often rely on predefined basis functions or parametric structures that lack the flexibility to capture complex nonlinear relationships. Additionally, strong assumptions about the underlying data distributions further limit their ability to achieve a data-driven orthogonal effect decomposition. To address these challenges, this study proposes a functional-output orthogonal additive Gaussian process (FOAGP) to efficiently perform the data-driven orthogonal effect decomposition. By enforcing a conditional orthogonality constraint on the separable prior process, the proposed functional-output orthogonal additive kernel enables data-driven orthogonality without requiring prior distributional assumptions. The FOAGP framework also provides analytical formulations for local Sobol' indices and expected conditional variance sensitivity indices, enabling comprehensive sensitivity analysis by capturing both global and local effect significance. Validation through two simulation studies and a real case study on fuselage shape control confirms the model's effectiveness in orthogonal effect decomposition and variance decomposition, demonstrating its practical value in engineering applications.

Industrial Internet of Things (IIoT) connects machines, and it is important to build intelligent models to prevent machine failures by identifying incipient faults. To develop intelligent fault diagnosis models, empirical risk minimization (ERM)-based modeling paradigm has been prevalently applied. However, during model training, ERM primarily focuses on instance-to-prototype (ItP) distances from a prototypical perspective, which may limit its effectiveness in analyzing data of diverse distributions. To improve the ERM model, we propose considering additional instance-to-instance distances (ItI) and prototype-to-prototype (PtP) distances, leading to a new modeling framework-distance-aware risk minimization (DARM). To gain awareness of extra types of distances, two novel losses are proposed based on reformulations of soft-max cross entropy. Theoretical explorations are conducted to justify the significance of collectively considering ItP, ItI, and PtP distances. Methodologically, DARM can jointly minimize three types of distance-aware losses to train neural networks for fault diagnosis in the same fashion as ERM. In a comprehensive computational study, DARM consistently outperformed ERM in domain generalization (DG) tasks based on various machine fault diagnosis data sets. In addition, DARM has superior performance over several recent DG methods. The code is available at https://github.com/mozhenling/doge-darm.

Portfolio management is one of the unresponded problems of the Motion Pictures Industry (MPI). To design an optimal portfolio for an MPI distributor, it is essential to predict the box office of each project. Moreover, for an accurate box office prediction, it is critical to consider the effect of the celebrities involved in each MPI project, which was impossible with any precedent expert-based method. Additionally, the asymmetric characteristic of MPI data decreases the performance of any predictive algorithm. In this paper, firstly, the fame score of the celebrities is determined using a large language model. Then, to tackle the asymmetric character of MPI's data, projects are classified. Furthermore, the box office prediction takes place for each class of projects. Finally, using a hybrid multi-attribute decision-making technique, the preferability of each project for the distributor is calculated, and benefiting from a bi-objective optimization model, the optimal portfolio is designed.

Accurate identification of community-dwelling older adults at high fall risk can facilitate timely intervention and significantly reduce fall incidents. Analyzing gait and balance capabilities via feature extraction and modeling through sensor-based motion data has emerged as a viable approach for fall risk assessment. However, the existing approaches for extracting key features related to fall risk lack inclusiveness, with limited consideration of the non-linear characteristics of sensor signals, such as signal complexity, self-similarity, and local stability. In this study, we developed a multifaceted feature extraction scheme employing diverse feature types, including demographic, descriptive statistical, non-linear, spatiotemporal and spectral features, derived from three-axis accelerometers and gyroscope data. This study is the first attempt to investigate non-linear features related to fall risk in multi-task scenarios from a dynamic system perspective. Based on the extracted multifaceted features, we propose an ensemble elastic net (E-E-N) approach for handling imbalanced data and offering high model interpretability. The E-E-N utilizes bootstrap sampling to construct base classifiers and employs a weighting mechanism to aggregate the base classifiers. We conducted a set of validation experiments using real-world data for comprehensive comparative analysis. The results demonstrate that the E-E-N approach exhibits superior predictive performance on fall risk classification. Our proposed approach offers a cost-effective tool for accurately assessing fall risk and alleviating the burden of continuous health monitoring in the long term.

Large-scale Gaussian process (GP) modeling is becoming increasingly important in machine learning. However, the standard modeling method of GPs, which uses the maximum likelihood method and the best linear unbiased predictor, is designed to run on a single computer, which often has limited computing power. Therefore, there is a growing demand for approximate alternatives, such as composite likelihood methods, that can take advantage of the power of multiple computers. However, these alternative methods in the literature offer limited options for practitioners because most methods focus more on computational efficiency rather than statistical efficiency. Limited accurate solutions to the parameter estimation and prediction for fast GP modeling are available in the literature for supercomputing practitioners. Therefore, this study develops an optimal composite likelihood (OCL) scheme for distributed GP modeling that can minimize information loss in parameter estimation and model prediction. The proposed predictor, called the best linear unbiased block predictor (BLUBP), has the minimum prediction variance given the partitioned data. Numerical examples illustrate that both the proposed composite likelihood estimation and prediction methods provide more accurate performance than their traditional counterparts under various cases, and an extremely close approximation to the standard modeling method is observed.

Fall, a leading cause of accidental death and injury in older adults aged 65 and above, has become a rapidly growing health concern in aging populations worldwide. Data-driven methods integrating depth imaging technology have received growing attention in automated fall risk assessment owing to their noninvasiveness and less dependence on healthcare professionals. However, most existing depth image data-based models neglect the inherent physiological and potential functional connections and lack sufficient real-world data validation. To fill the research gap, we developed a novel approach named multiscale skeletal transformer (MSS-Former), leveraging depth image technology and deep-learning models for effective fall risk prediction. Our contributions mainly consist of four parts. First, we introduced a multimodel output feature fusion transformer in fall risk prediction, enabling output merging and weighting from multiple model streams dynamically. Second, we developed an innovative scheme to construct interjoint skeletal topology, systematically focusing on joints' intrinsic physiological and potential functional connections. Third, we constructed a ResNet-FPN, greatly enhancing multiscale feature extraction capabilities. Fourth, we conducted a field study in a local hospital and performed a comprehensive validation of our developed approach. The comparison results show that our approach achieved outstanding predictive performance, surpassing state-of-the-art methods on the real-world data set, with accuracy, precision, recall, and F1 scores of 97.84%, 97.33%, 96.97%, and 96.92%, respectively. In practice, the proposed approach would be of great value in the timely identification for individuals at high fall risk and facilitate decision making to take appropriate interventions.

Prognostics and health management (PHM) has gotten considerable attention in the background of Industry 4.0. Battery PHM contributes to the reliable and safe operation of electric devices. Nevertheless, relevant reviews are still continuously updated over time. In this paper, we browsed extensive literature related to battery PHM from 2018 to 2023 and summarized advances in battery PHM field, including battery testing and public datasets, fault diagnosis and prediction methods, health status estimation and health management methods. The last topic includes state of health estimation methods, remaining useful life prediction methods and predictive maintenance methods. Each of these categories is introduced and discussed in details. Based on this survey, we accordingly discuss challenges left to battery PHM, and provide future research opportunities. This research systematically reviews recent research about battery PHM from the perspective of key PHM steps and provide some valuable prospects for researchers and practitioners.

Background: Artificial intelligence (AI)-based medical devices and digital health technologies, including medical sensors, wearable health trackers, telemedicine, mobile health (mHealth), large language models (LLMs), and digital care twins (DCTs), significantly influence the process of clinical decision support systems (CDSS) in healthcare and medical applications. However, given the complexity of medical decisions, it is crucial that results generated by AI tools not only be correct but also carefully evaluated, understandable, and explainable to endusers, especially clinicians. The lack of interpretability in communicating AI clinical decisions can lead to mistrust among decision-makers and a reluctance to use these technologies. Objective: This paper systematically reviews the processes and challenges associated with interpretable machine learning (IML) and explainable artificial intelligence (XAI) within the healthcare and medical domains. Its main goals are to examine the processes of IML and XAI, their related methods, applications, and the implementation challenges they pose in digital health interventions (DHIs), particularly from a quality control perspective, to help understand and improve communication between AI systems and clinicians. The IML process is categorized into pre-processing interpretability, interpretable modeling, and post-processing interpretability. This paper aims to foster a comprehensive understanding of the significance of a robust interpretability approach in clinical decision support systems (CDSS) by reviewing related experimental results. The goal is to provide future researchers with insights for creating clinician-AI tools that are more communicable in healthcare decision support systems and offer a deeper understanding of their challenges. Methods: Our research questions, eligibility criteria, and primary goals were proved using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline and the PICO (population, intervention, control, and outcomes) method. We systematically searched PubMed, Scopus, and Web of Science databases using sensitive and specific search strings. Subsequently, duplicate papers were removed using EndNote and Covidence. A two-phase selection process was then carried out on Covidence, starting with screening by title and abstract, followed by a full-text appraisal. The Meta Quality Appraisal Tool (MetaQAT) was used to assess the quality and risk of bias. Finally, a standardized data extraction tool was employed for reliable data mining. Results: The searches yielded 2,241 records, from which 555 duplicate papers were removed. During the title and abstract screening step, 958 papers were excluded, and the full-text review step excluded 482 studies. Subsequently, in quality and risk of bias assessment, 172 papers were removed. 74 publications were selected for data extraction, which formed 10 insightful reviews and 64 related experimental studies. Conclusion: The paper provides general definitions of explainable artificial intelligence (XAI) in the medical domain and introduces a framework for interpretability in clinical decision support systems structured across three levels. It explores XAI-related health applications within each tier of this framework, underpinned by a review of related experimental findings. Furthermore, the paper engages in a detailed discussion of quality assessment tools for evaluating XAI in intelligent health systems. It also presents a step-by-step roadmap for implementing XAI in clinical settings. To direct future research toward bridging current gaps, the paper examines the importance of XAI models from various angles and acknowledges their limitations.

Accurately estimating battery health status is crucial to ensure safe battery operation. The datadriven estimation methods proposed in current literature typically require complete sets of voltage, current, temperature data throughout battery charge/discharge process. However, obtaining such comprehensive data in real-world applications is often impractical. In this paper, we propose to estimate the health status of commercial lithium-ion batteries based on statistical features derived from voltage data within a specific voltage interval. Our emphasis on voltage data is grounded in the wealth of information voltage curves usually provide for battery prognostics. To implement our approach, the time-sampled battery data is first converted into voltagesampled data series via resampling using cubic splines. Then, a set of statistical features is constructed based on Delta Q(V) statistics and incremental capacity analysis, aiming to capture more effective voltage information from both cycle and discharge time dimensions. The extracted features are then studied considering battery degradation mechanisms, different battery materials, and the choice of voltage interval. The proposed methodology is tested on both lithium iron phosphate batteries and lithium nickel manganese cobalt oxide batteries, and experimental results demonstrate that the Delta Q(V) statistics perform more reliably and robustly than incremental capacity features. In addition, the mean, minimum, and variance are highly correlated with battery capacity, while higher-order moments such as skewness and kurtosis present insignificant impacts. Compared to estimation based on all discharge data, estimation based on partial voltage information yield very competitive results, with all root mean square errors less than 1%.

Online accurate battery state-of-health (SOH) estimation is crucial for ensuring safe and reliable operations of electric vehicles (EVs). Yet, such estimation problem remains a challenge in reality due to complex battery degradation behaviors and dynamic EV operations. This article proposes a novel deep learning-based framework, a bilateral-branched visual transformer with dilated self-attention (Bi-ViT-DSA), for online SOH estimation. The proposed framework considers partial charging segments during incomplete charging based on two mainstream charging modes, the multistage fast-charging (MSFC) and constant-current constant-voltage (CCCV) charging. To incorporate multitimescale battery aging dynamics into SOH estimation, a novel biparty input structure is developed to convey both inner cycle and intracycle degradation information from raw data. The proposed Bi-ViT-DSA is developed to learn multitimescale high-level latent features from the biparty input in parallel for SOH estimation. A dilated self-attention (DSA) mechanism is developed to reduce redundant operations in modeling. Computational studies are conducted on datasets of batteries under different chemistries and test conditions. Results validate the feasibility and robustness of the proposed method and its superior performance over a set of state-of-the-art benchmarks.

Machinery condition monitoring and fault diagnosis has attracted much attention because it is beneficial to reducing maintenance costs and improving industrial profits. Adaptive fault components extraction (AFCE) is the most crucial step for machinery fault diagnosis, and its core is statistical indices. Existing statistical indices including kurtosis and correlated kurtosis are empirical statistical indices (ESIs), and they cannot exactly quantify fault-related information in signals and distinguish fault components from interferential components. Thus, the ESIs might be easily affected by random impulsive noise, low-frequency components, etc. To solve this problem, a new statistical index named optimized weights spectrum based index (OWSI) is proposed in this article. The OWSI satisfies two good properties to guarantee exact quantification of fault components and effectively distinguish interferential components. Moreover, a new OWSI-based methodology is proposed to realize AFCE, and it can be implemented with signal decomposition algorithms such as variational mode decomposition without needing careful parameters tuning. Bearing and gear real-world fault signals are studied to verify the effectiveness of the proposed methodology. Results show that the proposed methodology is superior to ESI-based methods including classic fast kurtogram and newly developed feature mode decomposition.