While deep learning models have contributed to the advancement of sensor-based Human Activity Recognition (HAR), it usually requires large amounts of annotated sensor data to extract robust features. To alleviate the limitations of data annotation, contrastive learning has been applied to sensor-based HAR. One of the essential factors of contrastive learning is data augmentation, significantly impacting the performance of pre-training. However, current popular augmentation methods do not achieve competitive performance in contrastive learning for sensor-based HAR. Motivated by this issue, we propose a new sensor data augmentation method by resampling, which introduces variable domain information and simulates realistic activity data by varying the sampling frequency to maximize the coverage of the sampling space. The resampling augmentation method was evaluated in supervised learning and contrastive learning (SimCLRHAR and MoCoHAR). In the experiment, we use four datasets, UCI-HAR, MotionSense, USC-HAD, and MobiAct, using the mean F1-score as the evaluation metric for downstream tasks. The experiment results show that the resampling data augmentation outperforms all state-of-the-art augmentation methods in supervised learning and contrastive learning with a small amount of labeled data. The results also demonstrate that not all data augmentation methods have positive effects in contrastive learning frameworks.

Background: People with chronic pain (CP) sometimes report impaired cognitive function, including a deficit of attention, memory, executive planning, and information processing. However, the association between CP and cognitive decline was still not clear. Our study aimed to assess the association of CP as a risk factor with cognitive decline among adults.Methods: We included data from clinical studies. Publications were identified using a systematic search strategy from PubMed, Embase, and Cochrane Library databases from inception to October 10, 2020. We used the mean cognitive outcome data and the standard deviations from each group. The standardized mean difference (SMD) or odds ratio (OR), and 95% confidence intervals (CI) were performed for each cognitive decline outcome. I²-values were assessed to quantify the heterogeneities.Results: We included 37 studies with a total of 52,373 patients with CP and 80,434 healthy control participants. Because these studies used different evaluative methods, we analyzed these studies. The results showed CP was associated with cognitive decline when the short-form 36 health survey questionnaire (SF-36) mental component summary (SMD = −1.50, 95% CI = −2.19 to −0.81), the Montreal cognitive assessment (SMD = −1.11, 95% CI = −1.60 to −0.61), performance validity testing (SMD = 3.05, 95% CI = 1.74 to 4.37), or operation span (SMD = −1.83, 95% CI = −2.98 to −0.68) were used. However, we got opposite results when the studies using International Classification of Diseases and Related Health Problems classification (OR = 1.58, 95% CI = 0.97 to 2.56), the Mini-Mental State Examination (SMD = −0.42, 95% CI = −0.94 to 0.10; OR = 1.14, 95% CI = 0.91 to 1.42), and Repeatable Battery for the Assessment of Neuropsychological Status memory component (SMD = −0.06, 95% CI = −0.37 to 0.25).Conclusion: There may be an association between CP and the incidence of cognitive decline when some cognitive, evaluative methods were used, such as short-form 36 health survey questionnaire, Montreal cognitive assessment, performance validity testing, and operation span.

Background: Chronic disease incidence among the elderly is increasing, which is correlated with the acceleration of population aging. Evolving internet technologies may help prevent and provide interventions for chronic diseases in an accelerating aging process. However, the impact of daily internet use on the incidence of chronic diseases is not well understood. Objective: This study aims to investigate whether daily internet use by middle-aged and older adults may inhibit or promote the occurrence of chronic diseases. Methods: We included participants from the China Health and Retirement Longitudinal Study (CHARLS), a longitudinal survey of Chinese residents aged ≥45 years. We assessed 8-year data from wave 1 (June 2011-March 2012) to wave 4 (July-September 2018) in CHARLS. Data from wave 4 were used for a cross-sectional study, and data from all 4 waves were used for a longitudinal study. Self-reported data were used to track variables, including internet use, use frequency, and the incidence of different chronic diseases. Cox proportional hazards modeling was applied in the longitudinal study to examine the relationship between daily internet use and chronic diseases among middle-aged and older adults, while adjusting for sociodemographic characteristics and health behaviors. In addition, longitudinal data were used to analyze internet usage trends, and cross-sectional data were used to analyze the factors influencing internet use. Results: Among the 20,113 participants included in the longitudinal analyses, internet use increased significantly, from 2% to 12.3%, between 2011 and 2018. The adjusted model found statistically significant relationships between daily internet use and a lower incidence of the following chronic diseases: hypertension (hazard ratio [HR] 0.78, 95% CI 0.65-0.95, P=.01), chronic lung disease (HR 0.74, 95% CI 0.57-0.97, P=.03), stroke (HR 0.69, 95% CI 0.50-0.94, P=.02), digestive disease (HR 0.73, 95% CI 0.58-0.91, P=.005), memory-related disorders (HR 0.58, 95% CI 0.37-0.91, P=.02), arthritis or rheumatism (HR 0.60, 95% CI 0.48-0.76, P<.001), asthma (HR 0.52, 95% CI 0.33-0.84, P=.007), depression (HR 0.80, 95% CI 0.71-0.89, P<.001), and vision impairment (HR 0.83, 95% CI 0.74-0.93, P=.004). Moreover, our study also showed that with increasing frequency of internet use, the risk of some chronic diseases decreases. Conclusions: This study found that middle-aged and older adults who use the internet have a reduced risk of developing chronic diseases versus those who do not use the internet. The increasing prevalence of daily internet use among middle-aged and older adults may stimulate contemplation of the potential role of internet platforms in future research on chronic disease prevention.

Emerging evidence suggests that Toll-like receptors (TLRs) ligands pretreatment may play a vital role in the progress of myocardial ischemia/reperfusion (I/R) injury. As the ligand of TLR3, polyinosinic-polycytidylic acid (poly(I:C)), a synthetic double-stranded RNA, whether its preconditioning can exhibit a cardioprotective phenotype remains unknown. Here, we report the protective effect of poly(I:C) pretreatment in acute myocardial I/R injury by activating TLR3/PI3K/Akt signaling pathway. Poly(I:C) pretreatment leads to a significant reduction of infarct size, improvement of cardiac function, and downregulation of inflammatory cytokines and apoptotic molecules compared with controls. Subsequently, our data demonstrate that phosphorylation of TLR3 tyrosine residue and its interaction with PI3K is enhanced, and protein levels of phospho-PI3K and phospho-Akt are both increased after poly(I:C) pretreatment, while knock out of TLR3 suppresses the cardioprotection of poly(I:C) preconditioning through a decreased activation of PI3K/Akt signaling. Moreover, inhibition of p85 PI3K by the administration of LY294002 in vivo and knockdown of Akt by siRNA in vitro significantly abolish poly(I:C) preconditioning-induced cardioprotective effect. In conclusion, our results reveal that poly(I:C) preconditioning exhibits essential protection in myocardial I/R injury via its modulation of TLR3, and the downstream PI3K/Akt signaling, which may provide a potential pharmacologic target for perioperative cardioprotection.

The massive generation of time-series data by large-scale Internet of Things (IoT) devices necessitates the exploration of more effective models for multivariate time-series forecasting. In previous models, there was a predominant use of the channel dependence (CD) strategy (where each channel represents a univariate sequence). Current state-of-the-art (SOTA) models primarily rely on the channel independence (CI) strategy. The CI strategy treats channel multichannel series as separate single-channel series, expanding the data set to improve generalization performance and avoiding interchannel correlation that disrupts long-term features. However, the CI strategy faces the challenge of interchannel correlation forgetting. To address this issue, we propose an innovative Mixed Channels strategy, combining the data expansion advantages of the CI strategy with the ability to mitigate interchannel correlation forgetting. Based on this strategy, we introduce MCformer, a multivariate time-series forecasting model with mixed channel features. The model blends a specific number of channels, leveraging an attention mechanism to effectively capture interchannel correlation information when modeling long-term features. Experimental results demonstrate that the Mixed Channels strategy outperforms pure CI strategy in multivariate time-series forecasting tasks.

Wearable sensor-based human activity recognition (HAR) is a critical research domain in activity perception. However, achieving high efficiency and long sequence recognition remains a challenge. Despite the extensive investigation of temporal deep learning models, such as convolutional neural networks, RNNs, and transformers, their extensive parameters often pose significant computational and memory constraints, rendering them less suitable for resource-constrained mobile health applications. This study introduces HARMamba, an innovative lightweight and versatile HAR architecture that combines selective bidirectional state-space model and hardware-aware design. To optimize real-time resource consumption in practical scenarios, HARMamba employs linear recursive mechanisms and parameter discretization, allowing it to selectively focus on relevant input sequences while efficiently fusing scan and recompute operations. The model employs independent channels to process sensor data streams, dividing each channel into patches and appending classification tokens to the end of the sequence. It utilizes position embedding to represent the sequence order. The patch sequence is subsequently processed by HARMamba Block, and the classification head finally outputs the activity category. The HARMamba Block serves as the fundamental component of the HARMamba architecture, enabling the effective capture of more discriminative activity sequence features. HARMamba outperforms contemporary state-of-the-art frameworks, delivering comparable or better accuracy with significantly reducing computational and memory demands. Its effectiveness has been extensively validated on four publicly available data sets, namely, PAMAP2, WISDM, UNIMIB SHAR, and UCI. The F1 scores of HARMamba on the four data sets are 99.74%, 99.20%, 88.23%, and 97.01%, respectively.