Browse Articles
Discover research articles across all indexed journals
Evaluating augmented reality in physical education for dyslexic students from the perspectives of teachers and students
The landscape of renal protein S-acylation in mice with lipid-induced nephrotoxicity
The effect of video-based multimedia information before amniocentesis on pain, anxiety, and pregnancy outcomes
Elucidating the chrononutrition patterns and sleep quality among subfertile patients with different vitamin D levels
Construction and validation of a nomogram based on the log odds of positive lymph nodes to predict the prognosis of T1 gastric cancer
Efficient and assured reinforcement learning-based building HVAC control with heterogeneous expert-guided training
Abstract Building heating, ventilation, and air conditioning (HVAC) systems account for nearly half of building energy consumption and $$20\%$$ of total energy consumption in the US. Their operation is also crucial for ensuring the physical and mental health of building occupants. Compared with traditional model-based HVAC control methods, the recent model-free deep reinforcement learning (DRL) based methods have shown good performance while do not require the development of detailed and costly physical models. However, these model-free DRL approaches often suffer from long training time to reach a good performance, which is a major obstacle for their practical deployment. In this work, we present a systematic approach to accelerate online reinforcement learning for HVAC control by taking full advantage of the knowledge from domain experts in various forms. Specifically, the algorithm stages include learning expert functions from existing abstract physical models and from historical data via offline reinforcement learning, integrating the expert functions with rule-based guidelines, conducting training guided by the integrated expert function and performing policy initialization from distilled expert function. Moreover, to ensure that the learned DRL-based HVAC controller can effectively keep room temperature within the comfortable range for occupants, we design a runtime shielding framework to reduce the temperature violation rate and incorporate the learned controller into it. Experimental results demonstrate up to 8.8X speedup in DRL training from our approach over previous methods, with low temperature violation rate.
Microbiota shifts in fracture-related infections and pathogenic transitions identified by 16S rDNA sequencing
Association between chronic diseases and depressive inclinations among rural middle-aged and older adults
Study on lightweight strategies for L-YOLO algorithm in road object detection
Operation mechanism analysis and parameter optimization of airflow-rotating disc separation device for agricultural film fragments
A study of the effect of motor structural components on harmonic noise
Development of a PANoptosis-related LncRNAs for prognosis predicting and immune infiltration characterization of gastric Cancer
Nickel silicide nanowire anodes for microbial fuel cells to advance power production and charge transfer efficiency in 3D configurations
High-efficiency stepdown/step-up converter for series-connected energy storage system
Blood predictive biomarkers for cognitive impairment among community-dwelling older adults: a cross-sectional study in China
Self-supervised learning reduces label noise in sharp wave ripple classification
Abstract In the field of electrophysiological signal analysis, the classification of time-series datasets is essential. However, these datasets are often compromised by the prevalent issue of incorrect attribution of labels, known as label noise, which may arise due to insufficient information, inappropriate assumptions, specialists’ mistakes, and subjectivity, among others. This critically impairs the accuracy and reliability of data classification, presenting significant barriers to extracting meaningful insights. Addressing this challenge, our study innovatively applies self-supervised learning (SSL) for the classification of sharp wave ripples (SWRs), high-frequency oscillations involved in memory processing that were generated before or after the encoding of spatial information. This novel SSL methodology diverges from traditional label correction techniques. By utilizing SSL, we effectively relabel SWR data, leveraging the inherent structural patterns within time-series data to improve label quality without relying on external labeling. The application of SSL to SWR datasets has yielded a 10% increase in classification accuracy. While this improved classification accuracy does not directly enhance our understanding of SWRs, it opens up new pathways for research. The study’s findings suggest the transformative capability of SSL in improving data quality across various domains reliant on precise time-series data classification.
Performance assessment of disposable carbon-based immunosensors for the detection of SARS-CoV-2 infections
Abstract We designed, developed, and clinically tested two rapid antigen-based immunosensors for SARS-CoV-2 detection, enabling diagnosis and viral load quantification for under USD $2. In a first clinical study, a screen-printed disposable carbon-based (SPC) sensor was assessed on prospectively recruited adult participants classified into three study groups: healthy donors (n = 46); SARS-CoV-2-infected symptomatic patients (n = 58); and co-habitants of patients without prior testing (n = 38). Nasopharyngeal aspirates (NA), oropharyngeal swabs (OS), and saliva (SA) samples were obtained from all participants. Performance was measured in terms of clinical sensitivity and specificity against a reference diagnostic RT-qPCR kit and analytical sensitivity (limit of detection, LoD) and specificity using recombinant material in lab tests. A second study was performed using the same sensor design, albeit with laser-induced graphene (LIG) electrodes, using nasopharyngeal swabs (NS) on 224 patient samples obtained at different stages of the pandemic, of which 110 tested negative and 114 positive via RT-qPCR. We find OS was the most informative sample, when compared to NA and SA. The SPC-based sensors had a 93.8% sensitivity and 61.5% specificity with OS samples, while the LIG-based sensors with NS had a lower sensitivity of 68.93%, albeit a significantly higher specificity of 86.17%. We believe specificity values for the SPC sensors were driven by positive results from co-habitants and healthy donors and were affected by the low sensitivity (75.5%) and high LoD (> 20,000 viral copies/mL) of the reference RT-qPCR kit used, and the lower sensitivity of the LIG-based was due to a reduced set of effective antigen-binding sites caused by the non-covalent LIG-mAb ligands used. The immunosensor’s LoD to spike protein in phosphate-buffered saline (PBS) for both types of sensors was near 1 fg/mL and showed no cross-reactivity to recombinant structural proteins of Epstein-Barr and Influenza. Performance metrics and time-to-result (5 < 12 min) provide proof-of-principle of the immunosensor’s applicability as a low-cost, rapid technology for determining SARS-CoV-2 infections. Changing the working electrode material to LIG, instead of SPC, improved specificity even in the presence of pathogen variants. Discordant results between our two immunosensor versions and RT-qPCR tests are attributed not only to limited antibody effectiveness in the former but also to the quality of RT-qPCR probes used at the height of the pandemic.