Search

Virtual health has been acclaimed as a transformative force in healthcare delivery. Yet, its dropout issue is critical that leads to poor health outcomes, increased health, societal, and economic costs. Timely prediction of patient dropout enables stakeholders to take proactive steps to address patients' concerns, potentially improving retention rates. In virtual health, the information asymmetries inherent in its delivery format, between different stakeholders, and across different healthcare delivery systems hinder the performance of existing predictive methods. To resolve those information asymmetries, we propose a Multimodal Dynamic Knowledge-driven Dropout Prediction (MDKDP) framework that learns implicit and explicit knowledge from doctor-patient dialogues and the dynamic and complex networks of various stakeholders in both online and offline healthcare delivery systems. We evaluate MDKDP by partnering with one of the largest virtual health platforms in China. MDKDP improves the F1-score by 3.26 percentage points relative to the best benchmark. Comprehensive robustness analyses show that integrating stakeholder attributes, knowledge dynamics, and compact bilinear pooling significantly improves the performance. Our work provides significant implications for healthcare IT by revealing the value of mining relations and knowledge across different service modalities. Practically, MDKDP offers a novel design artifact for virtual health platforms in patient dropout management.

Alkyl radicals are powerful intermediates for the generation of carbon-carbon bonds, which play an indispensable role in the synthesis of natural products, pharmaceuticals, and pesticides. Traditionally, there are two main methods for the generation of alkyl radicals. The first is C-H bond functionalization via hydrogen-atom-transfer (HAT). HAT processes have been used as an effective approach for selectively activating C-H bonds via radical pathways. The other strategy to explore the generation of alkyl radicals is C-X bond functionalization via halogen-atom-transfer (XAT). Alkyl halides are one of the largest classes of building blocks in synthesis and they can be obtained from the corresponding alcohols. The most straightforward and effective way to form such alkyl radicals is the direct homolytic cleavage of C-X bonds. In past decades, photoredox catalysis has emerged as a powerful and greener tool for the synthesis of radicals under mild reaction conditions, which has brought tremendous attention. Although remarkable success has been made in this field, some methods still require costly transition metal catalysts or toxic reagents. Herein, we display a series of visible light-induced approaches under transition-metal free conditions or using earth-abundant metals. These novel photo-induced transformations and corresponding mechanistic work will be discussed in the following order: We will first present our work on metal-free visible-light-promoted C(sp3)-H functionalization of aliphatic cyclic ethers using trace O2. This reaction uses a trace amount of aerobic oxygen as the sole green oxidant under blue light at room temperature to achieve the synthesis of sulfone and phosphate derivatives in good to excellent yields using cyclic ethers and vinyl sulfones. Then, we report on a photo-induced C(sp3)-H chalcogenation of amide derivatives and ethers via a ligand-to-metal charge-transfer. This reaction converts secondary and tertiary amides, sulfonamides, and carbamates into the corresponding amido-N,S-acetal derivatives in good yields, using an earth abundant metal catalyst under mild conditions. Finally, we present a photoredox polyfluoroarylation of alkyl halides via halogen atom transfer. This method converts primary, secondary, and tertiary unactivated abundant alkyl halides into the corresponding polyfluoroaryl compounds in good yields and has good functional group compatibility.

Contemporary organizations recognize the importance of lean and green production to realize ecological and economic benefits. Compared with the existing optimization methods, the multi-task multi-objective reinforcement learning (MT-MORL) offers an attractive means to address the dynamic, multi-target process-optimization problems associated with Energy-Flexible Machining (EFM). Despite the recent advances in reinforcement learning, the realization of an accurate Pareto frontier representation remains a major challenge. This article presents a generative manifold-based policy-search method to approximate the continuously distributed Pareto frontier for EFM optimization. To this end, multi-pass operations are formulated as part of a multi-policy Markov decision process, wherein the machining configurations witness dynamic changes. However, the traditional Gaussian distribution cannot accurately fit complex upper-level policies. Thus, a multi-layered generator was designed to map the high-dimensional policy manifold from a simple Gaussian distribution without performing complex calculations. Additionally, a hybrid multi-task training approach is proposed to handle the mode collapse and large task difference observed during the improvement of the generalization performance. Extensive computational testing and comparisons against existing baseline methods have been performed to demonstrate the improved Pareto frontier quality and computational efficiency of the proposed algorithm.

Contemporary organizations recognize the importance of lean and green production to realize ecological and economic benefits. Compared to the existing optimization methods, the multi-task multi-objective reinforcement learning (MT-MORL) offers an attractive means to address the dynamic, multi-target process-optimization problems associated with energy-flexible machining (EFM). Despite the recent advances in RL, the realization of an accurate Pareto frontier representation remains a major challenge. This paper presents a generative manifold-based policy-search method to approximate the continuously distributed Pareto frontier for EFM optimization. To this end, multi-pass operations are formulated as part of a multi-policy Markov decision process, wherein the machining configurations witness dynamic changes. However, the traditional Gaussian distribution cannot accurately fit complex upper-level policies. Thus, a multi-layered generator was designed to map the high-dimensional policy manifold from a simple Gaussian distribution without performing complex calculations. Additionally, a hybrid multi-task training approach is proposed to handle the mode collapse and large task difference observed during the improvement of the generalization performance. Extensive computational testing and comparisons against existing baseline methods and four meta-heuristics have been performed to demonstrate the improved Pareto frontier quality and computational efficiency of the proposed algorithm.

Supplemental material, sj-docx-1-wmr-10.1177_0734242X211036254 for Predictors of waste sorting and recycling behavioural intention among youths: Evidence from Shenzhen, China and Turku, Finland by Miaojia Huang, Kris M. Y. Law, Shuang Geng, Ben Niu and Pekka Kettunen in Waste Management & Research

Supplemental material, sj-docx-1-wmr-10.1177_0734242X211036254 for Predictors of waste sorting and recycling behavioural intention among youths: Evidence from Shenzhen, China and Turku, Finland by Miaojia Huang, Kris M. Y. Law, Shuang Geng, Ben Niu and Pekka Kettunen in Waste Management & Research

In the Internet of things (IoT), the flow and sharing of data between different devices has become a trend. However, determining how to ensure the controlled flow and sharing of data is an...

The starve of optimized solutions through Artificial Intelligence and broadening its scope in machine health monitoring through diversified machine learning methods got significance from last two decades. Fault identification in complex engineering systems and its pattern recognition has always been challenging phase for plant crew. Online condition monitoring of plant by dedicated observers, maintaining huge volume of plant running data through operating indicators, acquiring signals through multiple sensors, plant management tools using signal processing techniques, selection and transfer of useful data to a field-oriented expert is a big chaos. This paper overcome these dreads using Artificial Intelligence that provides a quick real time condition monitoring and fault detection ability and robustness in decision making with high diagnostic accuracy. The paper equates the existing fault diagnosis techniques with machine learning. Supervised Artificial Neural Network and Convolutional Neural Network have applied to detect machine anomalies and found to be efficient methods for fast data processing, identification and classification of normal and abnormal condition of machine. Supervised AI, data driven machine health monitoring methods proposed in the paper have shown tremendous results for handling and classification of different faults from massive amount of plant machinery data.

Swarm optimization algorithms and agent based modeling (ABM) are two closely related research areas, parts of the multi agent system field, but they are traditionally not combined. Swarm optimization, in this case the bacterial foraging optimization (BFO), searches for an optimal solution while the ABM searches for a conclusion which resembles the real world, and it can be far from optimal. To bridge the gap, the overall goal this paper is to propose a new paradigm in the form of an architecture and operation procedures, thus creating a BFO-ABM hybrid. The other goal is to create a method which enables 3D visualization of the BFO algorithm. Firstly, an environment is created together with bacteria which physically perform all operators of the BFO. Secondly, a way of seamlessly embedding the bacteria from the BFO into the ABM environment is described. The bacteria are then manipulated and motivated with food and toxicity to act in a certain agent-like way. Simulation results prove that the agents can be effectively used as an ABM tool to present agents of all sizes and behaviors resembling numerous things, from companies, vehicles to people.

Traditionally, there are two models on differential privacy: the central model and the local model. The central model focuses on the machine learning model and the local model focuses on the training data. In this paper, we study the \textit{input perturbation} method in differentially private empirical risk minimization (DP-ERM), preserving privacy of the central model. By adding noise to the original training data and training with the `perturbed data', we achieve ($\epsilon$,$\delta$)-differential privacy on the final model, along with some kind of privacy on the original data. We observe that there is an interesting connection between the local model and the central model: the perturbation on the original data causes the perturbation on the gradient, and finally the model parameters. This observation means that our method builds a bridge between local and central model, protecting the data, the gradient and the model simultaneously, which is more superior than previous central methods. Detailed theoretical analysis and experiments show that our method achieves almost the same (or even better) performance as some of the best previous central methods with more protections on privacy, which is an attractive result. Moreover, we extend our method to a more general case: the loss function satisfies the Polyak-Lojasiewicz condition, which is more general than strong convexity, the constraint on the loss function in most previous work.

Informative features play a crucial role in the single image super-resolution task. Channel attention has been demonstrated to be effective for preserving information-rich features in each layer. However, channel attention treats each convolution layer as a separate process that misses the correlation among different layers. To address this problem, we propose a new holistic attention network (HAN), which consists of a layer attention module (LAM) and a channel-spatial attention module (CSAM), to model the holistic interdependencies among layers, channels, and positions. Specifically, the proposed LAM adaptively emphasizes hierarchical features by considering correlations among layers. Meanwhile, CSAM learns the confidence at all the positions of each channel to selectively capture more informative features. Extensive experiments demonstrate that the proposed HAN performs favorably against the state-of-the-art single image super-resolution approaches., Comment: 16 pages, 6 figures, IEEE International Conference on Computer Vision

Inter-annual variability in total precipitation can lead to significant changes in carbon flux. In this study, we used the eddy covariance (EC) technique to measure the net CO2 ecosystem exchange (NEE) of an alpine meadow in the northern Tibetan Plateau. In 2005 the meadow had precipitation of 489.9 mm and in 2006 precipitation of 241.1 mm, which, respectively, represent normal and dry years as compared to the mean annual precipitation of 476 mm. The EC measured NEE was 87.70 g C m-2 yr-1 in 2006 and –2.35 g C m-2 yr-1 in 2005. Therefore, the grassland was carbon neutral to the atmosphere in the normal year, while it was a carbon source in the dry year, indicating this ecosystem will become a CO2 source if climate warming results in more drought conditions. The drought conditions in the dry year limited gross ecosystem CO2 exchange (GEE), leaf area index (LAI) and the duration of ecosystem carbon uptake. During the peak of growing season the maximum daily rate of NEE and Pmax and α were approxim...

We are interested in low-dimensional dynamics in an ensemble of coupled nonidentical generalized oscillators on the 3-sphere. The system of governing equations for such an ensemble is referred to as non-Abelian Kuramoto model in the literature. We establish an analogue (or an extension) of the Ott-Antonsen (OA) result for this model., Comment: One proposition is incorrect

Enclosure is one of the most widely used management tools for degraded alpine grassland on the northern Tibetan Plateau, but the responses of different types of grassland to enclosure may vary, and research on these responses can provide a scientific basis for improving ecological conservation. This study took one site for each of three grassland types (alpine meadow, alpine steppe and alpine desert) on the northern Tibetan Plateau as examples, and explored the effects of enclosure on plant and soil nutrients by comparing differences in plant community biomass, leaf-soil nutrient content and their stoichiometry between samples from inside and outside the fence. The results showed that enclosure can significantly increase all aboveground biomass in these three grassland types, but it only increased the 10–20 cm underground biomass in the alpine desert. Enclosure also significantly increased the leaf nutrient content of the dominant plants and contents of total nitrogen (N), total potassium (K), and organic carbon (C) in 10–20 cm soil in alpine desert, thus changing the stoichiometry between C, N and P (phosphorus). However, enclosure significantly increased only the N content of dominant plant leaves in alpine steppe, while other nutrients and stoichiometries of both plant leaves and soil did not show significant differences in alpine meadow and alpine steppe. These results suggested that enclosure has differential effects on these three types of alpine grasslands on the northern Tibetan Plateau, and the alpine desert showed the most active ecological conservation in the responses of its soil and plant nutrients.

Quantifying the relationship between the drought severity index and climate factors is crucial for predicting drought risk in situations characterized by climate change. However, variations in drought risk are not readily discernible under conditions of climate change, and this is particularly the case on the Tibetan Plateau. This study examines the correlations between the annual drought severity index (DSI) and 14 climate factors (including temperature, precipitation, humidity, wind speed, and hours of sunshine factors), on the Tibetan Plateau from 2000 to 2011. Spatial average DSI increased with precipitation and minimum relative humidity, while it decreased as the hours of sunshine increased. The correlation between DSI and climate factors varied with vegetation types. In alpine meadows, the correlation of the spatial DSI average with the percentage of sunshine and hours of sunshine (P

Livestock grazing is one of primary way to use grasslands throughout the world, and the forage-livestock balance of grasslands is a core issue determining animal husbandry sustainability. However, there are few methods for assessing the forage-livestock balance and none of those consider the dynamics of external abiotic factors that influence forage yields. In this study, we combine long-term field observations with remote sensing data and meteorological records of temperature and precipitation to quantify the impacts of climate change and human activities on the forage-livestock balance of alpine grasslands on the northern Tibetan Plateau for the years 2000 to 2016. We developed two methods: one is statical method based on equilibrium theory and the other is dynamic method based on non-equilibrium theory. We also examined the uncertainties and shortcomings of using these two methods as a basis for formulating policies for sustainable grassland management. Our results from the statical method showed severe overgrazing in the grasslands of all counties observed except Nyima (including Shuanghu) for the entire period from 2000 to 2016. In contrast, the results from the dynamic method showed overgrazing in only eight years of the study period 2000–2016, while in the other nine years alpine grasslands throughout the northern Tibetan Plateau were less grazed and had forage surpluses. Additionally, the dynamic method found that the alpine grasslands of counties in the northeastern and southwestern areas of the northern Tibetan Plateau were overgrazed, and that alpine grasslands in the central area of the plateau were less grazed with forage surpluses. The latter finding is consistent with field surveys. Therefore, we suggest that the dynamic method is more appropriate for assessment of forage-livestock management efforts in alpine grasslands on the northern Tibetan Plateau. However, the statical method is still recommended for assessments of alpine grasslands profoundly disturbed by irrational human activities.

Plant photosynthesis is the fundamental driver of all the biospheric functions. Alpine meadow on the Tibetan Plateau is sensitive to rapid climate change, and thus can be considered an indicator for the response of terrestrial ecosystems to climate change. However, seasonal variations in photosynthetic parameters, including the fraction of photosynthetically active radiation by canopy (FPAR), the light extinction coefficient (k) through canopy, and the leaf area index (LAI) of plant communities, are not known for alpine meadows on the Tibetan Plateau. In this study, we used field measurements of radiation components and canopy structure from 2009 to 2011 at a typical alpine meadow on the northern Tibetan Plateau to calculate these three photosynthetic parameters. We developed a satellite-based (NDVI and EVI) method derived from the Beer-Lambert law to estimate the seasonal dynamics of FPAR, k ,and LAI, and we compared these estimates with the Moderate Resolution Imaging Spectroradiometer (MODIS) FPAR (FPAR_MOD) and LAI product (LAI_MOD). The results showed that the average daily FPAR was 0.33, 0.37 and 0.35, respectively, from 2009 to 2011, and that the temporal variations could be explained by all four satellite-based FPAR estimations, including FPAR_MOD, an FPAR estimation derived from the Beer-Lambert law with a constant k (FPAR_LAI), and two FPAR estimations from the nonlinear functions between the ground measurements of FPAR (FAPRg) and NDVI/EVI (FPAR_NDVI and FPAR_EVI). We found that FPAR_MOD seriously undervalued FPARg by over 40%. Tower-based FPAR_LAI also significantly underestimated FPARg by approximately 20% due to the constant k (0.5) throughout the whole growing seasons. This indicated that using FPAR_LAI to validate the FPAR_MOD was not an appropriate method in this alpine meadow because the seasonal variation of k ranged from 0.19 to 2.95 in this alpine meadow. Thus, if the seasonal variation of k was taken into consideration, both FPAR_NDVI and FPAR_EVI provided better descriptions, with negligible overestimates of less than 5% of FAPRg (RMSE=0.05), in FPARg estimations than FPAR_MOD and FPAR_LAI. Combining the satellite-based (NDVI and EVI) estimations of seasonal FPAR and k, LAI_NDVI and LAI_EVI derived from the Beer-Lambert law also provided better LAIg estimations than LAI_MOD (less than 30% of LAIg). Therefore, this study concluded that satellite-based models derived from the Beer-Lambert law were a simple and efficient method for estimating the seasonal dynamics of FPAR, k and LAI in this alpine meadow.

Double Hopf bifurcation analysis can be used to reveal some complicated dynamical behavior in a dynamical system, such as the existence or coexistence of periodic orbits, quasi-periodic orbits, or even chaos. In this paper, an algorithm for deriving the normal form near a codimension-two double Hopf bifurcation of a reaction-diffusion system with time delay and Neumann boundary condition is rigorously established, by employing the center manifold reduction technique and the normal form method. We find that the dynamical behavior near bifurcation points are proved to be governed by twelve distinct unfolding systems. Two examples are performed to illustrate our results: for a stage-structured epidemic model, we find that double Hopf bifurcation appears when varying the diffusion rate and time delay, and two stable spatially inhomogeneous periodic oscillations are proved to coexist near the bifurcation point; in a diffusive predator-prey system, we theoretically proved that quasi-periodic orbits exist on two- or three-torus near a double Hopf bifurcation point, which will break down after slight perturbation, leaving the system a strange attractor.

We investigate a diffusive, stage-structured epidemic model with the maturation delay and freely-moving delay. Choosing delays and diffusive rates as bifurcation parameters, the only possible way to destabilize the endemic equilibrium is through Hopf bifurcation. The normal forms of Hopf bifurcations on the center manifold are calculated, and explicit formulae determining the criticality of bifurcations are derived. There are two different kinds of stable oscillations near the first bifurcation: on one hand, we theoretically prove that when the diffusion rate of infected immature individuals is sufficiently small or sufficiently large, the first branch of Hopf bifurcating solutions is always spatially homogeneous; on the other, fixing this diffusion rate at an appropriate size, stable oscillations with different spatial profiles are observed, and the conditions to guarantee the existence of such solutions are given. These phenomena are investigated by calculating the corresponding eigenfunction of the Laplacian at the first Hopf bifurcation point.

Supplemental Material, Online_Appendices_-_Niu_(1) for Equilibria and Location Choice in Corporate Tax Regimes by Ben J. Niu in Public Finance Review

With the rapid development of location-aware devices such as smart phones, Location-Based Services (LBSs) are becoming increasingly popular. Users can enjoy convenience by sending queries to LBS servers and obtaining service information that is nearby. However, these queries may leak the users' locations and interests to the un-trusted LBS servers, leading to serious privacy concerns. In this paper, we propose a Privacy-Preserving Pseudo-Location Updating System (3PLUS) to achieve k-anonymity for mobile users using LBSs. In 3PLUS, without relying on a third party, each user keeps pseudo-locations obtained from both the history locations and the encountered users, and randomly exchanges one of them with others when encounters occur. As a result, each user's buffer is disordered. A user can obtain any k locations from the buffer to achieve k-anonymity locally. The security analysis shows the security properties and our evaluation results indicate that the user's privacy is significantly improved.

We studied biological characteristics and habitat requirements of Parnassius imperator in Yagou,Yongjing County.We also analyzed population trend and key factors responsible for its fluctuation.Our research methods included for example transect survey,observation at sample plots,indoor breeding,etc. 1)P.imperator is an endemic species to China.It is univoltine in Yongjing County,and overwinters in egg stage on rocks nearby host plants Corydalis adunca(also endemic to China).Eggs hatch in mid-March the following year.The larvae stage includes 4 instars and last for 52 days.Duration of the larva period is from late March to late June.The pupae stage lasts about 47 days,and larvae pupate in early May.Adults emerge at the beginning of May and can fly fast.Peak quantity of adults ranges from mid-July to mid-August,however adults can still be found in late September.The egg stage lasts about eight months. 2)P.imperator is a dominant species in Yagou,Yongjing County.Adults like to stay and fly in valley with bare rocks and shrubs,larvae are found on sunny and semi-sunny slopes.Geographic distribution of P.imperator and its host plant C.adunca overlaps in China.This explains their close co-evolutionary relationship. 3)Two subspecies of P.imperator.P.imperator gigas Kotsch,occured in Yongjing county and P.imperator regulus(Bryk et Eisner) in Hezuo city. 4)Extremely abnormal climate,for example,Strong cooling,frost or snow,is the main factor that affects the population decline of P.imperator.Human disturbance is afnother threat affecting survival of P.imperator. 5)Conversation measures should focus on conservation of habitats of Parnassius sp.,and reduction of human disturbance to support survival of the population in Yagou.We also stress the need for long term population monitoring and research on impact of climate chang and habitat loss.

The transmission of 1.5 keV-electrons through a conical glass capillary is reported. This study aims to understand the so-called guiding effect for the negatively charged particles (e.g. electrons). The guiding mechanism is understood quite well with positively charged particles in particular highly charged ions, but not clear with electrons, i.e., even the basic scheme mediated by the existence of negative charge patches to guide the electrons is still somewhat controversial. The study of the charging-up dynamics causing the electrons transport inside the capillary will shed light on this issue. In order to perform this, a data acquisition system has been setup to follow the time evolution of the twodimensional angular distribution of the transmitted electrons. The electrons are detected by the multi-channel plate (MCP) detector with a phosphor screen. The image from the phosphor screen is recorded by a charge-coupled device camera. The timing signals for the detected events are extracted from the back stack of the MCP detector and recorded by the data acquisition system, synchronized with the acquired images. The electron beam has a size of 0.5 mm0.5 mm and a divergence of less than 0.35. The inner diameter of the straight part of the capillary is 1.2 mm and the exit diameter is 225 m. A small conducting aperture of 0.3 mm in diameter is placed at the entrance of the capillary. Two-dimensional angular distribution of the transmitted electrons through conical glass capillary and its time evolution are measured. The results show that the transmission rate decreases and reaches to a constant value for the completely discharged glass capillary with time going by. The centroid of the angular distribution moves to an asymptotic value while the width remains unchanged. These transmission characteristics are different from those indicated in our previous work (2016 Acta Phys. Sin. 65 204103). The difference originates from the different manipulations of the capillary outer surface. A conducting layer is coated on the outer surface of the capillary and grounded in this work. This isolates various discharge/charge channels and forms a new stable discharge channel. The transmission rate as a function of the tilt angle shows that the allowed transmission occurs at the tilt angle limited by the geometrical factors, i.e., the geometrical opening angle given by the aspect ratio as well as the beam divergence. The transmission characteristics suggest that most likely there are formed no negative patches to facilitate the electron transmission through the glass capillary at this selected beam energy. It is different from that of highly charged ions, where the formation of the charge patches prohibits the close collisions between the following ions and guides them out of the capillary.

The structure system of the steel roof of Tianjin Grand Theater is complex, and the stress distribution changes constantly in the unloading process. Through the simulation analysis of the whole unloading process, the stress distribution and deformation of the structure in different unloading steps were obtained. Besides, member configuration was predicated and preset in order to satisfy the target configuration when the project is completed. The simulation analysis verifies the feasibility of the unloading plan and provides a reference for the design and construction of long-span space steel structures in the future.

DNA methylation regulates the transcription of genes without changing their coding sequences. It plays a vital role in the process of embryogenesis and tumorgenesis. To gain more insights into how such epigenetic mechanism works in the human cells, we apply the two popular data mining techniques, i.e., Rough Sets, and Decision Trees, to uncover the logical rules of DNA methylation. Our results show that the Rough Sets method can generate and utilize fewer rules to fully separate the methylation dataset, whereas Decision Trees method relies on more rules but involves fewer decision variables to do the same task. We also find that some of the gene promoters are highly comethylated, demonstrating the evidence that genes are highly interactive epigenetically in human cells.

In this paper we propose the two dimensional Laplacianfaces method for face recognition. The new algorithm is developed based on the two techniques, i.e., locality preserved embedding and image based projection. The two dimensional Laplacianfaces method is not only computationally more efficient but also more accurate than the one dimensional Laplacianfaces method in extracting the facial features for human face authentication. Extensive experiments are performed to test and evaluate the new algorithm using the Yale and the AR face databases. The experimental results indicate that the two dimensional Laplacianfaces method significantly outperforms the existing two dimensional Eigenfaces, the two dimensional Fisherfaces and the one dimensional Laplacianfaces methods under the various settings of experiment conditions.