A SEASONAL IMPUTATION METHOD FOR ADDRESSING MISSING DATA IN ENVIRONMENTAL IOT SENSOR TIME SERIES

High-quality monthly upper-air temperature and humidity datasets of 22 stations have been developed for monitoring and assessing long-term trends in temperature and humidity over Australia. In addition, high quality monthly upper-air temperature and humidity time series have been developed for five Australian remote island sites: the subantarctic Macquarie Island, Lord Howe Island in the Tasman Sea, Norfolk Island in the southwest Pacific Ocean, Willis Island in the Coral Sea and Cocos Island in the eastern Indian Ocean. The datasets are based on 23UTC soundings with most records starting in 1958 for temperature and in 1965 for humidity. The quality control process involved the examination of station metadata (historical information about stations and recordings) and an objective statistical test which detected discontinuities in the data series. At each station and each level, data series were adjusted (on a monthly timescale) when discontinuities were identified. Homogenised temperature and humidity (represented by the dew point temperature, DWPT) data were analysed at mandatory levels. Temperature time series were found to be mostly complete since 1958, with higher levels having more missing data than lower levels. For humidity it was found that there were insufficient daily observations to compute monthly means in the period 1965 to 1991 at all levels except 850 hPa. Furthermore, DWPT data contained a number of biases due to changes in technology. To be able to analyse humidity data at least at the two lowest levels (850 and 700 hPa), we explored the possibility of removing part of the daily data, in order to make the record over the period 1965-2011 consistent (i.e. independent of technological changes). The all-Australian mean annual temperature shows statistically significant increases/decreases over the 1958-2011 period at all examined levels except the 150 hPa level. Since 1958 temperatures at the 850 and 700 hPa levels increased about 1.2°C, while at the 500, 400 and 300 hPa levels they increased about 1.6°C. At levels which intercept the tropopause height, temperature over the same period of time increased by 0.9°C at the 200 hPa and a decreased by -0.3°C at the 150 hPa. In the lower stratosphere, at the 100 hPa level, temperature decreased by -1.8°C. A similar vertical structure of changes in temperature is found in remote islands data. The observed patterns of change in the upper-air temperature over Australia are in agreement with physical principles and model simulations under the enhanced greenhouse effect, both of which show general warming in the troposphere and cooling in the lower stratosphere (noting that stratospheric cooling is predominantly due to the decrease in stratospheric ozone and to a smaller extent to the increase of greenhouse gases). Results are also consistent with the expectation that the troposphere should warm at a faster rate than the surface. Analysis of the homogenised humidity data indicates that over the 1965-2011 period, all-Australian DWPT shows statistically significant increases at the 850 and 700 hPa levels. The increase since 1965 at 850 hPa is about 0.9°C (comparable to the temperature tend), while at 700 hPa the increase in DWPT is nearly double in magnitude, about 1.rc (larger than the temperature trend). The increase in humidity in the lower troposphere is in accordance with the expectation that, as the troposphere warms, the amount of moisture in it should increase. It is also consistent with an upward trend in observed Australian near-surface humidity.

Throughout the transportation process from the factory to the end-user, many medicines and vaccines need to be kept under specific conditions to preserve their effectiveness. This involves passing through different phases like shaping time, storage time. Temperature, humidity, and light intensity are commonly utilized in a variety of industries nationwide, including agricultural research, food safety, pharmaceutical storage, the chemical industry, environmental protection, and other sectors. Environmental Data monitoring, also known as a temperature, humidity, and lux recorder, is primarily employed for monitoring and recording temperature, humidity, and light levels (lux) during the storage and transportation of food, medicine, and perishable goods. With the advancement of modern technology, temperature, humidity, and lux data loggers are increasingly crucial in our daily lives.

Improvement in anesthesia and surgical practices has enabled more patients, including those undergoing higher-risk surgeries, to be treated in outpatient settings. This shift creates a need for reliable postoperative monitoring at home. Wearable devices like the Biobeat® offer continuous, real-time monitoring of vital signs have remained largely untested for home use in this context.A prospective, single-center observational study was conducted at the Centre hospitalier de l'Université de Montréal (CHUM) from February to August 2023. Fifty eligible patients underwent continuous monitoring with the Biobeat® device for five days post-surgery, with data transmitted to CHUM's telehealth service. Feasibility was assessed by the percentage of patients without data loss during consecutive 2-hour intervals.Of the 50 patients enrolled, 49 completed the study, but all experienced some level of data loss. While 39.6% of patients maintained connectivity without complete data loss for 6-8-hour intervals, challenges included device discomfort, Bluetooth disconnection, and connectivity issues. Thirteen patients removed the device early due to discomfort or technical issues. Of the 3 patients who experienced post-operative complications, no data was available within 24h prior to the episodes. Continuous vital signs monitoring is feasible for high-risk outpatient surgery patients; however, significant improvements are required in device reliability and data accessibility. Further studies are needed to refine the technology and develop reliable protocols for postoperative monitoring in the home setting.

This data article presents a comprehensive maternal health risk dataset comprising 6058 samples (from an initially collected 6103 samples, after quality validation and outlier removal) collected through Internet of Things (IoT)-enabled Medical Cyber-Physical Systems (MCPS) across nine healthcare facilities in Bangladesh between February 2021 and January 2023. The dataset includes eight clinical features: age, body temperature, heart rate, systolic and diastolic blood pressure, body mass index (BMI), hemoglobin A1c (HbA1c), and fasting blood glucose levels, with expert- validated three-level risk classifications (high, mid, low). Data collection employed standardized IoT sensors with Raspberry Pi 4 controllers and medical-grade sensor arrays. Comprehensive quality validation removed 45 physio- logically implausible samples (0.74%), ensuring all values fall within clinically acceptable ranges. Comprehensive evaluation demonstrates dual-purpose utility(1) immediate clinical deployment for hos- pital decision support systems and (2) research baseline infrastructure for future temporal modeling studies-validated through rigorous experimental analysis. Traditional machine learning models achieved clinical-grade performance (defined as ≥95% accuracy with balanced precision/recall, consistent with established clinical decision support benchmarks), with XGBoost and Random Forest both reaching 99.34% accuracy (95% confidence interval: 98.84–99.75%), confirming exceptional data quality for immediate clinical deployment in hospital decision support systems. Feature importance analysis identified diastolic blood pressure (18.1%), body temperature (17.7%), and systolic blood pres- sure (17.0%) as primary predictive indicators. Additionally, recurrent neural network (RNN) models achieved 93.98% accuracy using synthetic temporal sequences (treating features as timesteps), establishing architectural baselines for future longitudinal studies. The 5.36% performance gap between XGBoost and Simple RNN (p < 0.001, McNemar’s test) confirms the dataset’s cross-sectional structure is optimally suited for ensemble learning methods, while RNN experiments provide proof-of-concept infrastructure for temporal modeling when genuine longitudinal measurements become available. This dataset addresses the critical need for maternal health risk prediction in resource-limited set- tings, supporting both immediate clinical deployment and future research directions. The data demonstrate readiness for integration into clinical decision support systems (validated by 99.34% accuracy), temporal modeling research infrastructure (93.98% RNN baseline), and healthcare technology development. All data underwent rigorous quality control with physiologically validated ranges, expert risk classification, and ethical approval (DUET-IRB-2021–034), making it suitable for academic research, clinical implementation, and public health policy formulation in developing countries.

A smart society is an empowered society, which can improve the lives of its citizens by using the latest innovations and technologies. This improvement can happen in several dimensions out of which security is a major one. Inconsistency and forgery are very common phenomenon where handwritten signatures are often preserved for training a classifier to authenticate a person. The removal of outliers, at the outset, obviously improves the quality of training and the classifier. The present article deals with the mechanized segregation of the poor-quality authentic signatures from reliable ones. Machine learning algorithms for outlier handling utilizing clustering, classification and statistical techniques have been implemented in this context. Subsequent performance evaluation after outlier removal reflects improvement of both true positive and true negative recognition rate accuracy. The performance evaluation presents the significant differences between authentication accuracy and forgery accuracy in the context of building a safe, secure and smart society.

3D object detection is a crucial technology in fields such as autonomous driving and robotics. As a direct representation of the 3D world, point cloud data plays a vital role in feature extraction and geometric representation. However, in real-world applications, point cloud data often suffers from occlusion, resulting in incomplete observations and degraded detection performance. Existing methods, such as PG-RCNN, generate semantic surface points within each Region of Interest (RoI) using a single grid size. However, a fixed grid scale cannot adequately capture multi-scale features. A grid that is too small may miss fine structures—especially problematic when dealing with small or sparse objects—while a grid that is too large may introduce excessive background noise, reducing the precision of feature representation. To address this issue, we propose an enhanced PG-RCNN architecture with a Multi-Scale Grid Attention Module as the core contribution. This module improves the expressiveness of point features by aggregating multi-scale information and dynamically weighting features from different grid resolutions. Using a simple linear transformation, we generate attention weights to guide the model to focus on regions that contribute more to object recognition, while effectively filtering out redundant noise. We evaluate our method on the KITTI 3D object detection validation set. Experimental results show that, compared to the original PG-RCNN, our approach improves performance on the Cyclist category by 2.66% and 2.54% in the Moderate and Hard settings, respectively. Additionally, our approach shows more stable performance on small object detection tasks, with an average improvement of 2.57%, validating the positive impact of the Multi-Scale Grid Attention Module on fine-grained geometric modeling, and highlighting the efficiency and generalizability of our model.

Utilizing machine learning techniques for data-driven diagnosis of high temperature PEM fuel cells is beneficial and meaningful to the system durability. Nevertheless, ensuring the robustness of diagnosis remains a critical and challenging task in real application. To enhance the robustness of diagnosis and achieve a more thorough evaluation of diagnostic performance, a robust diagnostic procedure based on electrochemical impedance spectroscopy (EIS) and a new method for evaluation of the diagnosis robustness was proposed and investigated in this work. To improve the diagnosis robustness: (1) the degradation mechanism of different faults in the high temperature PEM fuel cell was first analyzed via the distribution of relaxation time of EIS to determine the equivalent circuit model (ECM) with better interpretability, simplicity and accuracy; (2) the feature extraction was implemented on the identified parameters of the ECM and extra attention was paid to distinguishing between the long-term normal degradation and other faults; (3) a Siamese Network was adopted to get features with higher robustness in a new embedding. The diagnosis was conducted using 6 classic classification algorithms—support vector machine (SVM), K-nearest neighbor (KNN), logistic regression (LR), decision tree (DT), random forest (RF), and Naive Bayes employing a dataset comprising a total of 1935 collected EIS. To evaluate the robustness of trained models: (1) different levels of errors were added to the features for performance evaluation; (2) a robustness coefficient (Roubust_C) was defined for a quantified and explicit evaluation of the diagnosis robustness. The diagnostic models employing the proposed feature extraction method can not only achieve the higher performance of around 100% but also higher robustness for diagnosis models. Despite the initial performance being similar, the KNN demonstrated a superior robustness after feature selection and re-embedding by triplet-loss method, which suggests the necessity of robustness evaluation for the machine learning models and the effectiveness of the defined robustness coefficient. This work hopes to give new insights to the robust diagnosis of high temperature PEM fuel cells and more comprehensive performance evaluation of the data-driven method for diagnostic application.

Humans exhibit a remarkable ability to reliably classify sound sources in the environment even in presence of high levels of noise. In contrast, most engineering systems suffer a drastic drop in performance when speech signals are corrupted with channel or background distortions. Our brains are equipped with elaborate machinery for speech analysis and feature extraction, which hold great lessons for improving the performance of automatic speech processing systems under adverse conditions. The work presented here explores a biologically-motivated multi-resolution speaker information representation obtained by performing an intricate yet computationally-efficient analysis of the information-rich spectro-temporal attributes of the speech signal. We evaluate the proposed features in a speaker verification task performed on NIST SRE 2010 data. The biomimetic approach yields significant robustness in presence of non-stationary noise and reverberation, offering a new framework for deriving reliable features for speaker recognition and speech processing.

Simple SummaryBreast cancer is one of the leading causes of cancer death among women. Developing machine learning-based diagnosis models receives great attention from researchers and scientists using histopathology images. Deep learning (DL) algorithms automatically extract features from raw data through convolutional operations. The generalization of the DL models’ results relies on large datasets, although they eliminate the expert knowledge in the feature extraction stage. This work aimed to compare the performance of the features extracted via deep learning and a knowledge-based approach in breast cancer detection from histopathology images.Cancer develops when a single or a group of cells grows and spreads uncontrollably. Histopathology images are used in cancer diagnosis since they show tissue and cell structures under a microscope. Knowledge-based and deep learning-based computer-aided detection is an ongoing research field in cancer diagnosis using histopathology images. Feature extraction is vital in both approaches since the feature set is fed to a classifier and determines the performance. This paper evaluates three feature extraction methods and their performance in breast cancer diagnosis. Features are extracted by (1) a Convolutional Neural Network, (2) a transfer learning architecture VGG16, and (3) a knowledge-based system. The feature sets are tested by seven classifiers, including Neural Network (64 units), Random Forest, Multilayer Perceptron, Decision Tree, Support Vector Machines, K-Nearest Neighbors, and Narrow Neural Network (10 units) on the BreakHis 400× image dataset. The CNN achieved up to 85% for the Neural Network and Random Forest, the VGG16 method achieved up to 86% for the Neural Network, and the knowledge-based features achieved up to 98% for Neural Network, Random Forest, Multilayer Perceptron classifiers.

The FARMSCAPE approach to decision support: farmers', advisers', researchers' monitoring, simulation, communication and performance evaluation

The evolution of smart grid and the deepening of the power market have significantly enhanced the key position of precise electricity price estimation in industry decision‐making and the optimal resource allocation. It is necessary to implement short‐term electricity price forecast in the power market that accurately characterizes the dynamic change laws of electricity prices in a market environment with significant high volatility and nonlinear characteristics. To address this demand, the paper proposes an Attention‐based long short‐term memory model called ATT‐LSTM, which intends to improve the accuracy and interpretability of electricity price forecast, and provide support for power market operation analysis and trading decision making. In the model, multi‐dimensional feature engineering and data preprocessing are first performed, including meteorological data imputation, outlier removal, and time series sliding window construction. An attention layer is introduced to dynamically allocate importance weights to different time steps and features, allowing the model to adaptively focus on key influencing variables. This model is particularly suitable for electricity price forecast scenarios with significant seasonality and short‐term disturbances. Experiments are based on measured data from the Australian National Electricity Market, with comparisons conducted using multiple benchmark models, that is, LSTM, BPNN, and ATT‐LSTM. The results show that the ATT‐LSTM model achieves the optimal performance in terms of multiple metrics (e.g., MAE and RMSE) and prove that introducing the attention mechanism into the LSTM can effectively improve the accuracy and transparency of electricity price forecast.

Environmental monitoring data are collected at and around the Area 3 and Area 5 Radioactive Waste Management Sites (RWMSs) within the Nevada National Security Site (NNSS). These data include direct radiation exposure, as well as radiation from the air, groundwater, meteorology, and vadose zone. This report summarizes the 2015 environmental data to provide an overall evaluation of RWMS performance and to support environmental compliance and performance assessment (PA) activities. Some of these data (e.g., radiation exposure, air, and groundwater) are presented in other reports developed by National Security Technologies, LLC. Direct radiation monitoring data indicate exposure levels at the RWMSs are within the range of background levels measured at the NNSS. Slightly elevated exposure levels outside the Area 3 RWMS are attributed to nearby historical aboveground nuclear weapons tests. Air monitoring data show that tritium concentrations in water vapor and americium and plutonium concentrations in air particles are below Derived Concentration Standards for these radionuclides. Groundwater monitoring data indicate the groundwater in the uppermost aquifer beneath the Area 5 RWMS is not impacted by RWMS operations. Results of groundwater analysis from wells around the Area 5 RWMS were all below established investigation levels. Leachate samples collected from the leachate collection system at the mixed low-level waste cell were below established contaminant regulatory limits. During 2015, precipitation at the Area 3 RWMS was 0.9% above average, and precipitation at the Area 5 RWMS was 25% above average. Water balance measurements indicate that evapotranspiration from the vegetated weighing lysimeter dries the soil and prevents downward percolation of precipitation more effectively than evaporation as measured from the bare-soil weighing lysimeter. The 1.8 inches of precipitation in September reached the lowest sensors at 180 cm on the Cell 5S and 5N covers, however the sensors on the floor do not indicate any signs of the precipitation percolating thru the waste to the floor. Moisture from precipitation did not percolate below 150 centimeters (cm) (5 feet [ft]) in the vegetated final cover on the U-3ax/bl disposal unit at the Area 3 RWMS. A second remedial revegetation trial was done on the southern portion of the North Cover. In March 2015 it was determined that germination occurred on both the mulched and non-mulched treatments that were irrigated, but that no germination occurred on the non-irrigated treatments. By June the seedlings observed in March were gone, possibly grazed by rabbits which were observed inside the rabbit-proof fence. In August the entire area was sprayed with the herbicide glyphosate in an attempt to minimize weed seed production. During 2015, there was no drainage through 2.4 meters (8 ft) of soil as indicated from the Area 3 drainage lysimeters that received only natural precipitation. Thirty-one percent of the applied precipitation and irrigation drained from the bare-soil drainage lysimeter, which received 3 times the natural precipitation. All 2015 monitoring data indicate that the Area 3 and Area 5 RWMSs are performing within expectations of the model and parameter assumptions for the facilities’ PAs.

Environmental monitoring data were collected at and around the Area 3 and Area 5 Radioactive Waste Management Sites (RWMSs) at the Nevada National Security Site (NNSS). These data are associated with radiation exposure, air, groundwater, meteorology, vadose zone, subsidence, and biota. This report summarizes the 2010 environmental data to provide an overall evaluation of RWMS performance and to support environmental compliance and performance assessment (PA) activities. Some of these data (e.g., radiation exposure, air, and groundwater) are presented in other reports (National Security Technologies, LLC, 2010a; 2010b; 2011). Direct radiation monitoring data indicate exposure levels at the RWMSs are within the range of background levels measured at the NNSS. Air monitoring data at the Area 3 and Area 5 RWMSs indicate that tritium concentrations are slightly above background levels. All gamma spectroscopy results for air particulates collected at the Area 3 and Area 5 RWMS were below the minimum detectable concentrations, and concentrations of americium and plutonium are only slightly above detection limits. The measured levels of radionuclides in air particulates and moisture are below derived concentration guides for these radionuclides. Groundwater monitoring data indicate that the groundwater in the uppermost aquifer beneath the Area 5 RWMS is not impacted by facility operations. The 246.9 millimeters (mm) (9.72 inches [in.]) of precipitation at the Area 3 RWMS during 2010 is 56 percent above the average of 158.7 mm (6.25 in.), and the 190.4 mm (7.50 in.) of precipitation at the Area 5 RWMS during 2010 is 50 percent above the average of 126.7 mm (4.99 in.). Soil-gas tritium monitoring at borehole GCD-05 continues to show slow subsurface migration consistent with previous results. Water balance measurements indicate that evapotranspiration from the vegetated weighing lysimeter dries the soil and prevents downward percolation of precipitation more effectively than evaporation from the bare-soil weighing lysimeter. Data from the automated vadose zone monitoring system for the operational waste pit covers show that moisture from precipitation did not percolate below 90 centimeters (cm) (3 feet [ft]) before being removed by evaporation. Moisture from precipitation did not percolate below 61 cm (2 ft) in the vegetated final mono-layer cover on the U-3ax/bl disposal unit at the Area 3 RWMS before being removed by evapotranspiration. During 2010, there was no drainage through 2.4 meters (8 ft) of soil from the Area 3 drainage lysimeters that received only natural precipitation. Water drained from both the bare-soil drainage lysimeter and the invader species drainage lysimeter that received 3 times natural precipitation. All 2010 monitoring data indicate that the Area 3 and Area 5 RWMSs are performing within expectations of the model and parameter assumptions for the facility PAs.

Environmental monitoring data were collected at and around the Area 3 and Area 5 Radioactive Waste Management Sites (RWMSs) at the Nevada Test Site. These data are associated with radiation exposure, air, groundwater, meteorology, vadose zone, subsidence, and biota. This report summarizes the 2007 environmental data to provide an overall evaluation of RWMS performance and to support environmental compliance and performance assessment (PA) activities. Some of these data (e.g., radiation exposure, air, and groundwater) are presented in other reports (National Security Technologies, LLC, 2007a; 2008; Warren and Grossman, 2008). Direct radiation monitoring data indicate exposure levels at the RWMSs are at background levels. Air monitoring data at the Area 3 and Area 5 RWMSs indicate that tritium concentrations are slightly above background levels. A single gamma spectroscopy measurement for cesium was slightly above the minimum detectable concentration, and concentrations of americium and plutonium are only slightly above detection limits at the Area 3 RWMS. The measured levels of radionuclides in air particulates are below derived concentration guides for these radionuclides. Radon flux from waste covers is well below regulatory limits. Groundwater monitoring data indicate that the groundwater in the uppermost aquifer beneath the Area 5 RWMS is not impacted by facility operations. The 136.8 millimeters (mm) (5.39 inches [in.]) of precipitation at the Area 3 RWMS during 2007 is 13 percent below the average of 158.1 mm (6.22 in.), and the 123.8 mm (4.87 in.) of precipitation at the Area 5 RWMS during 2007 is 6 percent below the average of 130.7 mm (5.15 in.). Soil-gas tritium monitoring at borehole GCD-05U continues to show slow subsurface migration consistent with previous results. Water balance measurements indicate that evapotranspiration from the vegetated weighing lysimeter dries the soil and prevents downward movement percolation of precipitation more effectively than evaporation from the bare-soil weighing lysimeter. Data from the automated vadose zone monitoring system for the operational waste pit covers show that evaporation continues to slowly remove soil moisture that came from the heavy precipitation in the fall of 2004 and the spring of 2005. The vegetated final mono-layer cover on the U-3ax/bl disposal unit at the Area 3 RWMS effectively removes moisture from the cover by evapotranspiration. During 2007, there was no drainage through 2.4 meters (8 feet) of soil from the Area 3 drainage lysimeters that received only natural precipitation or were vegetated but water drained from the bare-soil Area 3 drainage lysimeter that received 3 times natural precipitation. Elevated tritium levels in plants and animals sampled from the Area 3 and Area 5 RWMSs show tritium uptake by the biota, but the low levels of other radionuclides do not suggest that there has been intrusion into the waste. All 2007 monitoring data indicate that the Area 3 and Area 5 RWMSs are performing within expectations of the model and parameter assumptions for the facility PAs.

Environmental monitoring data were collected at and around the Area 3 and Area 5 Radioactive Waste Management Sites (RWMSs) at the Nevada Test Site. These data are associated with radiation exposure, air, groundwater, meteorology, vadose zone, subsidence, and biota. This report summarizes the 2006 environmental data to provide an overall evaluation of RWMS performance and to support environmental compliance and performance assessment (PA) activities. Some of these data (e.g., radiation exposure, air, and groundwater) are presented in other reports (U.S. Department of Energy, 2006; Warren and Grossman, 2007; National Security Technologies, LLC, 2007). Direct radiation monitoring data indicate that exposure levels around the RWMSs are at or below background levels. Air monitoring data at the Area 3 and Area 5 RWMSs indicate that tritium concentrations are slightly above background levels. There is no detectable man-made radioactivity by gamma spectroscopy, and concentrations of americium and plutonium are only slightly above detection limits at the Area 3 RWMS. Measurements at the Area 5 RWMS show that radon flux from waste covers is no higher than natural radon flux from undisturbed soil in Area 5. Groundwater monitoring data indicate that the groundwater in the uppermost aquifer beneath the Area 5 RWMS is not impacted by facility operations. Precipitation during 2006 totaled 98.6 millimeters (mm) (3.9 inches [in.]) at the Area 3 RWMS and 80.7 mm (3.2 in.) at the Area 5 RWMS. Soil-gas tritium monitoring continues to show slow subsurface migration consistent with previous results. Moisture from precipitation at Area 5 remains at the bottom of the bare-soil weighing lysimeter, but this same moisture has been removed from the vegetated weighing lysimeter by evapotranspiration. Vadose zone data from the operational waste pit covers show that evaporation continues to slowly remove soil moisture that came from the heavy precipitation in the fall of 2004 and the spring of 2005. The vegetated final cover at U-3ax/bl continues to remove moisture by evapotranspiration. There was no drainage through 2.4 meters (8 feet) of soil from the Area 3 drainage lysimeters that received only natural precipitation or were vegetated. Water drained from the bare-soil Area 3 drainage lysimeter that received three times natural precipitation. All 2006 monitoring data indicate that the Area 3 and Area 5 RWMSs are performing within expectations of the model and parameter assumptions for the facility PAs.