Temporal Behavioral Patterns Research Articles

A Shared Economy Data Prediction Model Based on Deep Learning

In the dynamic and intricate shared economy, efficient resource management and forecasting are still crucial. In this research, a new prediction model is presented that aims to improve the operational efficiency of several shared economy services. Our method creatively combines Long Short-Term Memory (LSTM) networks with Genetic Algorithms (GA) to assess user demand and optimize resource deployment. We apply this methodology especially to e-scooter sharing services, but the underlying ideas and methods can be applied to other shared economy platforms as well, like peer-to-peer lending services, car sharing, and vacation rentals. The model starts with a GA to adjust the hyperparameters of the LSTM network, making the network better suited to handle specific characteristics of common economic data. Capturing the complex temporal and spatial patterns of user behavior and demand on these platforms requires this optimization. The LSTM element then predicts changes in service demand due to its capacity to analyze sequential records. Further to being useful for analyzing sequential data and predicting destiny wishes, this predictive functionality is important for a shared economy platform to correctly manage stock, allocate assets, and predict personal wishes. We use a large dataset to check our technique, demonstrating the predictive accuracy of the model and demand and its potential to aid strategic choice-making. Compared to traditional fashions, the consequences show a large development in forecast accuracy and resource allocation efficiency. Our methodology creates a robust basis for statistics-driven insights that decorate customer happiness and decorate the long-term increase of the shared financial system. This work highlights the blended ability of GA and LSTM inside the shared economy and paves the manner for future enhancements in using modern-day gadget mastering techniques to optimize and alter various shared services. In quick, effective useful resource control and forecasting in the shared economic system is tough, however our specific forecasting model combining GA-LSTM gives a manner ahead. Our technique, as it should be, predicts fluctuations in provider demand; it was first refined the usage of GA to regulate the LSTM hyperparameters. The consequences show how correct and powerful our version is and spotlight how it can enhance customer pleasure and operational performance. This research paves the manner for future trends within the software of gadgets, gaining knowledge of methods and supports the continuing enlargement and development of shared economic system offerings.

Quantitative index for temporal and spatial patterns of occupant behavior based on VRF big data

The energy consumption required to cool buildings has increased significantly because of occupant behavior. The air conditioning (AC) usage in residential buildings, particularly the part-time and part-space patterns, has a considerable impact on energy consumption. However, the part-time, part-space patterns of occupant behavior, which are characterized by complexity and diversity, present significant challenges when quantitatively assessing their impact on cooling energy consumption. To accurately describe occupant behavior, this study introduces the quantitative index of the part-time, part-space ratio of AC usage. We conducted clustering analysis based on part-time and part-space ratios of AC usage to obtain the typical occupant behavior patterns, user distribution, and cooling energy consumption levels of variable refrigerant flow (VRF) AC systems in various climatic zones of China based on data from 4373 VRF systems across various climatic zones from June to October 2019. The main findings were that the typical occupant behavior pattern of AC usage for residential buildings was the part-time, part-space AC mode, with less than 500 cooling hours and a simultaneous use coefficient below 0.6. The part-time, part-space ratio of AC usage effectively correlated occupant behavior with cooling energy consumption, as evidenced by an R2 value of 0.91, compared with the traditional standard of using the duration of the AC operation. The part-time, part-space ratio of AC usage and cooling energy consumption were higher in the hot summer and warm winter regions than those in the other two climatic zones. These findings can assist in the evaluation of cooling energy consumption for various occupant behavior patterns and provide technical guidance for reducing cooling energy consumption.

Retweeting behavior prediction based on dynamic Bayesian network classifier in microblogging networks

Nowadays, predicting user retweeting behavior in microblogging networks has gained considerable attention. Solving this problem allows for understanding the underlying mechanism of information diffusion and analyzing diffusion's evolution concerning a particular original microblog over time. Time-series data from microblogging networks are ubiquitous, and user retweeting behavior prediction is primarily considered a classification task. Consequently, there is a significant demand for accurate classification of time-series data. Several influential factors affect user retweeting behavior prediction in microblogging networks. However, much of the existing research focuses on studies that neglect the impact of users' indirect social influence strength. This neglect leads to inaccurate forecasts of indirect retweeting behavior for a particular original microblog. To address this, this paper presents a new time-sensitive measure of social influence strength based on a combination of users' temporal behavior patterns and the shortest cascade path length. Second, it proposes a prediction model based on Dynamic Bayesian Network derivative classifiers called V-DBNC to accurately predict user retweeting behavior over time. The proposed time-sensitive social influence strength factor and other factors, such as individual behavior and similarity-based driving factors, are applied to the V-DBNC model to achieve accurate user retweeting behavior prediction in the microblogging network. The joint density of factors in this classifier is estimated using a multivariate Gaussian kernel function with smoothing parameters. The V-DBNC model is empirically optimized by splitting the smoothing parameters into different-scale intervals and using the model averaging method to select the optimal classifier. Additionally, the proposed model's relatively simple structure helps overcome the overfitting problem and allows it to accumulate classification information through iterative evolution, promoting generalization. The real Twitter microblogging dataset is used to evaluate the performance of V-DBNC. The experimental results demonstrated that the proposed model outperforms other compared approaches when dealing with the classification of time-series data.

Quantifying Variability in Zebrafish Larvae Locomotor Behavior across Experimental Conditions: A Learning-Based Tracker

This study investigated the effects of environmental changes on zebrafish larval behavior, using single-factor and orthogonal experiments to assess locomotion during temperature and pH changes. In single-factor experiments, zebrafish larvae were exposed to variations in temperature (22 to 30 °C) and pH levels (6.0, 7.0, 9.0). The simultaneous temperature and pH changes were investigated by orthogonal tests. In both experiments, each zebrafish larva was recorded in three 5 min videos at different stages (before exposure, during short-term exposure (10 min), and after long-term exposure (60 min)). You Look Only Once (YOLOv5) and Deep Simple Online Real Time Tracking (DeepSORT) models were adopted to develop a zebrafish larva tracking system, and YOLOv5 was improved in two aspects of anchor clustering and network structure. The tracking accuracy of the tracking system for small targets effectively improved, reaching more than 98% MOTA (Multiple Object Tracking Accuracy). Principal Component Analysis (PCA) was employed to extract three behavioral features from 13 motion parameters, namely motion activity, edge behavior, and motion direction preference. Our findings reveal that lower temperatures and acidic conditions both led to a decrease in motion behavioral activity, and the former also increased edge behavior. Conversely, elevated temperatures and alkaline conditions had a muted impact on these behaviors. Interestingly, concurrent changes in temperature and pH significantly altered directional preference. Additionally, we observed that lower temperatures elicited distinct temporal behavioral patterns at a constant pH level. In summary, we recommend the precise control and explicit reporting of ambient temperature and pH in both breeding devices and experimental wells to minimize the environmental impact on zebrafish behavior and enhance experiment repeatability and reliability.

Spatial prey availability and pulsed reproductive tactics: Encounter risk in a canid-ungulate system.

Predation risk is a function of spatiotemporal overlap between predator and prey, as well as behavioural responses during encounters. Dynamic factors (e.g. group size, prey availability and animal movement or state) affect risk, but rarely are integrated in risk assessments. Our work targets a system where predation risk is fundamentally linked to temporal patterns in prey abundance and behaviour. For neonatal ungulate prey, risk is defined within a short temporal window during which the pulse in parturition, increasing movement capacity with age and antipredation tactics have the potential to mediate risk. In our coyote-mule deer (Canis latrans-Odocoileus hemionus) system, leveraging GPS data collected from both predator and prey, we tested expectations of shared enemy and reproductive risk hypotheses. We asked two questions regarding risk: (A) How does primary and alternative prey habitat, predator and prey activity, and reproductive tactics (e.g. birth synchrony and maternal defence) influence the vulnerability of a neonate encountering a predator? (B) How do the same factors affect behaviour by predators relative to the time before and after an encounter? Despite increased selection for mule deer and intensified search behaviour by coyotes during the peak in mule deer parturition, mule deer were afforded protection from predation via predator swamping, experiencing reduced per-capita encounter risk when most neonates were born. Mule deer occupying rabbit habitat (Sylvilagus spp.; coyote's primary prey) experienced the greatest risk of encounter but the availability of rabbit habitat did not affect predator behaviour during encounters. Encounter risk increased in areas with greater availability of mule deer habitat: coyotes shifted their behaviour relative to deer habitat, and the pulse in mule deer parturition and movement of neonatal deer during encounters elicited increased speed and tortuosity by coyotes. In addition to the spatial distribution of prey, temporal patterns in prey availability and animal behavioural state were fundamental in defining risk. Our work reveals the nuanced consequences of pulsed availability on predation risk for alternative prey, whereby responses by predators to sudden resource availability, the lasting effects of diversionary prey and inherent antipredation tactics ultimately dictate risk.

Where does the time go? Temporal patterns of pumping behaviors in mothers of very preterm infants vary by sociodemographic and clinical factors.

Mothers of very preterm (<32 weeks gestational age [GA]) infants are breast pump dependent and have shorter duration of milk provision than mothers of term infants. The opportunity (i.e., time) cost of pumping and transporting mother's own milk (MOM) from home to the NICU may be a barrier. There is a paucity of data regarding how much time mothers actually spend pumping. To investigate the variation in pumping behavior by postpartum week, maternal characteristics, and infant GA. Prospectively collected pump log data from mothers enrolled in ReDiMOM (Reducing Disparity in Mother's Own Milk) randomized, controlled trial included pumping date and start time and end time of each pumping session for the first 10 weeks postpartum or until the infant was discharged from the NICU, whichever occurred first. Outcomes included number of daily pumping sessions, number of minutes spent pumping per day, and pumping behaviors during 24-h periods, aggregated to the postpartum week. Medians (interquartile ranges) were used to describe outcomes overall, and by maternal characteristics and infant GA. Data included 13,994 pump sessions from 75 mothers. Maternal characteristics included 55% Black, 35% Hispanic, and 11% White and 44% <30 years old. The majority (56%) of infants were born at GA 28-31 weeks. Mothers pumped an average of less than 4 times per day, peaking in postpartum week 2. After accounting for mothers who stopped pumping, there was a gradual decrease in daily pumping minutes between postpartum weeks 2 (89 min) and 10 (46 min). Black mothers pumped fewer times daily than non-Black mothers after the first 2 weeks postpartum. On average mothers pumped less intensively than the minimum recommendation of 8 times and 100 min per day. However, these pumping behaviors represent significant maternal opportunity costs that should be valued by the institution and society at large.

Social behavior among nocturnally migrating birds revealed by automated moonwatching

Abstract Migrating birds often fly in group formations during the daytime, whereas at night, it is generally presumed that they fly singly. However, it is difficult to quantify group behavior during nocturnal migration as there are few means of directly observing interactions among individuals. We employed an automated form of moonwatching to estimate percentages of birds that appear to migrate in groups during the night within the Central Flyway of North America. We compared percentages of birds in groups across the spring and fall and examined overnight temporal patterns of group behavior. We found groups were rare in both seasons, never exceeding 10% of birds observed, and were almost nonexistent during the fall. We also observed an overnight pattern of group behavior in the spring wherein groups were more commonly detected early in the night and again just before migration activity ceased. This finding may be related to changes in species composition of migrants throughout the night, or alternatively, it suggests that group formation may be associated with flocking activity on the ground as groups are most prevalent when birds begin and end a night of migration.

The Application of Immersive Virtual Reality and Machine Learning for the Assessment of Unilateral Spatial Neglect

Abstract Even subtle forms of hemispatial neglect after stroke negatively affect the performance of daily life tasks, increase the risk of injury, and are associated with poor rehabilitation outcomes. Conventional paper-and-pencil tests, however, often underestimate the symptoms. We aimed to identify relevant neglect-specific measures and clinical decision rules based on machine learning techniques on behavioral data generated in a new Virtual Reality (VR) application, the immersive virtual road-crossing task. In total, 59 participants were included in our study: two right-hemispheric stroke groups with left neglect (N = 20) or no neglect (N = 19), classified based on conventional tests and medical diagnosis, and healthy controls (N = 20). A neuropsychological test battery and the VR task were administered to all participants. We applied decision trees and random forest models to predict the respective groups based on the results of the VR task. Our feature selection procedure yielded six features as suitable predictors, most of which involved lateral time-related measures, particularly reaction times, and head movements. Our model achieved a high training accuracy of 96.6% and estimated test accuracy of 76.8%. These results confirm previous reports that temporal behavioral patterns are key to detecting subtle neglect in patients with chronic stroke. Our results indicate that VR combined with machine learning has the potential to achieve higher test accuracies while being highly applicable to clinical practice.

Climate change influences the risk of physically harmful human-wildlife interactions

Interactions with wildlife can pose substantial physical risk to humans, as well as damage efforts to protect the species involved. News reports of increasing dangerous interactions with animals indicate climate change may be acting as a risk magnifier for these confrontations, yet its impacts on human-wildlife interactions remain uncertain in the scientific discourse. We analysed 331 media reports on climate change driven human wildlife conflicts involving physically dangerous species and verified the effects found in the media with evidence from scientific literature. Our analysis showed that climate change can increase wildlife-induced physical harm to people. This result was consistent for venomous species, terrestrial-and aquatic carnivorous species and large-bodied terrestrial animals in 44 countries across the globe. We identified four climate driven trends that impact the risk of human casualties: i) increased resource competition between humans and wildlife due to drought; ii) range expansion of dangerous animals due to higher average temperatures; iii) temporary displacement of wildlife due to extreme weather events; iv) and changes in temporal behaviour patterns of wildlife due to higher average temperatures. The identification of overarching trends across different regions and species show the need to bridge the gap between wildlife research and the study of climate-related risks. The existence of, or potential for, wildlife-induced physical harm to people should be taken into account as a component of climate change driven risk. At the same time, we stress the importance of including current and future impacts of climate change into long-term wildlife management- and conservation schemes.

MINT: Detecting Fraudulent Behaviors from Time-Series Relational Data

The e-commerce platforms, such as Shopee, have accumulated a huge volume of time-series relational data, which contains useful information on differentiating fraud users from benign users. Existing fraud behavior detection approaches typically model the time-series data with a vanilla Recurrent Neural Network (RNN) or combine the whole sequence as a single intention without considering the temporal behavioral patterns, row-level interactions, and different view intentions. In this paper, we present MINT, a M ultiview row- IN teractive T ime-aware framework to detect fraudulent behaviors from time-series structured data. The key idea of MINT is to build a time-aware behavior graph for each user's time-series relational data with each row represented as an action node. We utilize the user's temporal information to construct three different graph convolutional matrices for hierarchically learning the user's intentions from different views, that is, short-term, medium-term, and long-term intentions. To capture more meaningful row-level interactions and alleviate the over-smoothing issue in a vanilla time-aware behavior graph, we propose a novel gated neighbor interaction mechanism to calibrate the aggregated information by each action node. Since the receptive fields of the three graph convolutional layers are designed to grow nearly exponentially, our MINT requires many fewer layers than traditional deep graph neural networks (GNNs) to capture multi-hop neighboring information, and avoids recurrent feedforward propagation, thus leading to higher training efficiency and scalability. Our extensive experiments on the large-scale e-commerce datasets from Shopee with up to 4.6 billion records and a public dataset from Amazon show that MINT achieves superior performance over 10 state-of-the-art models and provides better interpretability and scalability.

Seasonal and circadian patterns of herring gull (Larus smithsoniansus) movements reveal temporal shifts in industry and coastal island interaction

AbstractAnthropogenic food subsidies attract opportunistic generalists like gulls in high densities, which may lead to negative impacts on human communities and local ecosystems. Managing impacts requires understanding why gulls use particular natural or industrial sites at different times of day or phases of the breeding cycle. Use of natural and human‐influenced habitats likely varies temporally as gulls alter schedules and site selection to match the predictability of different resources as they vary through space and time relative to patterns in human activities (seasonal industries, the work week, working hours) and natural rhythms (daylight, tide cycles), while gull resource requirements and restrictions to movement also shift with changing reproductive demands.We quantified seasonal and circadian patterns in American herring gull (Larus smithsoniansus) interactions with anthropogenic and natural sites throughout breeding using GPS data from 15 gulls tracked over 3 years from two colonies. We examined the weekly probability of gull occurrence at distinct destinations (e.g. breeding colony, islands, offshore, fish processing plants), and how occurrence varied with time of day, weekday/weekend, tide phase and colony, using GLMMs with a binomial response for destination‐specific occurrence.Probability at the colony varied predictably through the breeding season (highest attendance from dusk to dawn, during incubation and early chick rearing), providing confidence in the modelling approach for detecting temporal patterns in behaviour. Gulls visited other islands mostly outside incubation and early chick rearing, and from dusk through the night, likely roosting. Occurrence offshore where interaction with fishing vessels is possible was highest from dusk to dawn, and differed among colonies, but was the most likely destination during incubation and early chick rearing. Occurrence at fish plants gradually increased until after fledging when attendance was highest from Aug‐Oct coincident with the peak of Atlantic herring (Clupea harengus) processing and was more likely during the weekdays, during working hours, and during low and flood tide.Gulls in southwest Nova Scotia, Canada, have the behavioural flexibility to adapt to both natural rhythms and human schedules when beneficial, enabling them to thrive in a region where industry and natural resources are abundant. These findings can provide information to guide when and where to test different subsidy management strategies locally, while also considering potential increased pressures on island ecosystems. We emphasise that management outcomes of reductions of food subsidies for opportunistic species depend on multiple factors, including availability of alternative food sources and timing of use.

Privacy-Preserved Evolutionary Graph Modeling via Gromov-Wasserstein Autoregression

Real-world graphs like social networks are often evolutionary over time, whose observations at different timestamps lead to graph sequences. Modeling such evolutionary graphs is important for many applications, but solving this problem often requires the correspondence between the graphs at different timestamps, which may leak private node information, e.g., the temporal behavior patterns of the nodes. We proposed a Gromov-Wasserstein Autoregressive (GWAR) model to capture the generative mechanisms of evolutionary graphs, which does not require the correspondence information and thus preserves the privacy of the graphs' nodes. This model consists of two autoregressions, predicting the number of nodes and the probabilities of nodes and edges, respectively. The model takes observed graphs as its input and predicts future graphs via solving a joint graph alignment and merging task. This task leads to a fused Gromov-Wasserstein (FGW) barycenter problem, in which we approximate the alignment of the graphs based on a novel inductive fused Gromov-Wasserstein (IFGW) distance. The IFGW distance is parameterized by neural networks and can be learned under mild assumptions, thus, we can infer the FGW barycenters without iterative optimization and predict future graphs efficiently. Experiments show that our GWAR achieves encouraging performance in modeling evolutionary graphs in privacy-preserving scenarios.

Retention time and fix acquisition rate of glued-on GPS transmitters in a semi-aquatic species

BackgroundMovement is a core mechanism through which animals interact with their environment. GPS telemetry is a popular approach used to investigate animal movement, providing access to both the spatial and temporal behavioral patterns exhibited by an individual or population. However, while some species are easily tracked through traditional GPS attachment methods (such as GPS collars or backpacks), other species such as the North American beaver (Castor canadensis) present unique challenges given their fusiform shape and tapered neck.ResultsWe tested three different GPS transmitter attachment methods (tail-mounted, lower back glued-on, and upper back glued-on) for beavers over two seasons to determine which treatment was most effective in terms of retention time (RT, total number of days a transmitter remains attached) and GPS fix success rate (FSR, % of successful fixes vs. attempted) and investigated to what degree various factors (season, sex, and age class) affected these results. We then evaluated whether the data collected were sufficient for identifying home-ranging behavior (when an individual begins to display restricted space use and range residency). We found transmitters attached to the lower back during the fall to be the top performing treatment, having a similar mean FSR (51.59%) to upper back attachments in fall, but a significantly greater average RT (42.8 days). Of the 23 individuals included in the home-ranging behavior analysis, all but two had sufficient data for identifying home-ranging behavior.ConclusionsOur tests show that glued-on GPS tags can provide up to 2 months of fine-scale relocation data in a safe and effective manner. This allows the opportunity to answer novel questions regarding movement patterns of beavers and other semi-aquatic mammals.

Abstract MP69: Day-of-the-Week and Time-of-the-Day Temporal Patterns of Sedentary Behavior Among Hispanic / Latino Adults: Hispanic Community Health Study / Study of Latinos

Objectives: Existing sedentary behavior interventions have largely achieved modest results. Conventionally, interventions have attempted to reduce sedentary time using a full-day approach. An alternative strategy is to target specific periods when sedentary behavior is most prevalent. However, little is known when sedentary time is most likely to occur in adults. This study examined the day-of-the-week and time-of-the-day patterns of sedentary behavior among U.S. Hispanic/Latino adults. Methods: Participants (n=12,241) from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL), a multi-site prospective cohort study of Hispanic/Latino adults, were studied. Sedentary time was measured for one week using a hip-mounted accelerometer. Three sedentary metrics were quantified: total sedentary time, sedentary time in bouts ≥ 60 minutes, and number of sedentary breaks. Results: Temporal patterns across days of the week were statistically significant for all three sedentary metrics (Figure, Panels A-C, p &lt; .001). Total sedentary time decreased from 719±2 mins/d on Mondays to 710±3 mins/d on Fridays the lowest and increased over weekends with Sundays (733±3 mins/d) the highest. Temporal patterns across 3-h periods of the day were statistically significant for all sedentary metrics (Figure, Panels D-F, p &lt; .001). Time-of-the-day patterns had a U-curve pattern wherein adults were most sedentary late at night, became less sedentary throughout the day, reached peak activeness around noon, then became more sedentary into the evening. Day-of-the-week and time-of-the-day patterns were robust across age, employment, work shift, and acculturation subgroups. Conclusions: Mornings, evenings, and weekends were the most sedentary periods in this cohort of Hispanic/Latino adults, characterized by higher volumes of sedentary time, more prolonged sedentary bouts, and fewer sedentary breaks. These time periods should be considered as important windows for behavioral interventions targeting sedentary behavior.

Temporal learning analytics to explore traces of self-regulated learning behaviors and their associations with learning performance, cognitive load, and student engagement in an asynchronous online course.

Self-regulated learning (SRL) plays a critical role in asynchronous online courses. In recent years, attention has been focused on identifying student subgroups with different patterns of online SRL behaviors and comparing their learning performance. However, there is limited research leveraging traces of SRL behaviors to detect student subgroups and examine the subgroup differences in cognitive load and student engagement. The current study tracked the engagement of 101 graduate students with SRL-enabling tools integrated into an asynchronous online course. According to the recorded SRL behaviors, this study identified two distinct student subgroups, using sequence analysis and cluster analysis: high SRL (H-SRL) and low SRL (L-SRL) groups. The H-SRL group showed lower extraneous cognitive load and higher learning performance, germane cognitive load, and cognitive engagement than the L-SRL group did. Additionally, this study articulated and compared temporal patterns of online SRL behaviors between the student subgroups combining lag sequential analysis and epistemic network analysis. The results revealed that both groups followed three phases of self-regulation but performed off-task behaviors. Additionally, the H-SRL group preferred activating mastery learning goals to improve ethical knowledge, whereas the L-SRL group preferred choosing performance-avoidance learning goals to pass the unit tests. The H-SRL group invested more in time management and notetaking, whereas the L-SRL group engaged more in surface learning approaches. This study offers researchers both theoretical and methodological insights. Additionally, our research findings help inform practitioners about how to design and deploy personalized SRL interventions in asynchronous online courses.

Mining Implicit Behavioral Patterns via Attention Networks for Sequential Recommendation

Sequential recommender systems (SRSs) seek to model users’ dynamic preferences based on their interaction sequences to suggest items they may be interested in. Existing methods suffer from two major flaws in modeling user interest representations. First, most methods only encode the absolute positions of items in an interaction sequence, but ignore the relative position information between items, impeding the modeling of temporal patterns of user behaviors. At the same time, the multi-layer stacked network structure will cause the dilution of position information. Second, historical interactions usually contain many items that are irrelevant to the user's subsequent choices, while existing work lacks approaches to explicitly distinguish them from interest-related items. In this work, we alleviate the foregoing issues by proposing a user behavioral patterns enhanced attention network (BPERec), where a novel position encoding module enables the enhancement of temporal patterns implicit in interaction sequences. Additionally, an explicit denoising attention mechanism is proposed to enhance user interest representations. Experimental results on three benchmark datasets reveal that our method outperforms state-of-the-art baseline methods.

A Layered, Hybrid Machine Learning Analytic Workflow for Mouse Risk Assessment Behavior.

Accurate and efficient quantification of animal behavior facilitates the understanding of the brain. An emerging approach within machine learning (ML) field is to combine multiple ML-based algorithms to quantify animal behavior. These so-called hybrid models have emerged because of limitations associated with supervised [e.g., random forest (RF)] and unsupervised [e.g., hidden Markov model (HMM)] ML models. For example, RF models lack temporal information across video frames, and HMM latent states are often difficult to interpret. We sought to develop a hybrid model, and did so in the context of a study of mouse risk assessment behavior. We used DeepLabCut to estimate the positions of mouse body parts. Positional features were calculated using DeepLabCut outputs and were used to train RF and HMM models with equal number of states, separately. The per-frame predictions from RF and HMM models were then passed to a second HMM model layer ("reHMM"). The outputs of the reHMM layer showed improved interpretability over the initial HMM output. Finally, we combined predictions from RF and HMM models with selected positional features to train a third HMM model ("reHMM+"). This reHMM+ layered hybrid model unveiled distinctive temporal and human-interpretable behavioral patterns. We applied this workflow to investigate risk assessment to trimethylthiazoline and snake feces odor, finding unique behavioral patterns to each that were separable from attractive and neutral stimuli. We conclude that this layered, hybrid ML workflow represents a balanced approach for improving the depth and reliability of ML classifiers in chemosensory and other behavioral contexts.

Gap in the conditioned stimulus: Differential impacts on temporal expectancy in appetitive and aversive conditions in rats

We analyzed, through a Pavlovian conditioning procedure in rats, the temporal pattern of behavior in appetitive and aversive conditions within subjects, and the difference in inferred temporal working memory functioning with the Gap paradigm. For both conditions, we paired a 60-s conditioned stimulus (CS: tone1 or tone2) with an unconditioned stimulus (US: shock or chocolate pellet) delivered 20s after CS onset. The analyses of mean response rate and individual-trial data were performed during Probe trials, consisting of CS alone, and trials in which gaps of different position or duration were inserted, to assess the effect of the temporal manipulation on behavior. The results showed: (1) An anticipatory peak time in the aversive condition but better accuracy in the appetitive condition, (2) constancy in the Weber fraction suggesting that the difference in peak time was under clock control, (3) a graded effect of gap parameters only in the aversive condition and (4) different gap effects between conditions when a gap was inserted early in the CS. These results highlight behavioral differences between aversive and appetitive conditions and suggest that the temporal working memory mechanism was not engaged in the same manner in each condition.

Comprehensive space–time hydrometeorological simulations for estimating very rare floods at multiple sites in a large river basin

Abstract. Estimates for rare to very rare floods are limited by the relatively short streamflow records available. Often, pragmatic conversion factors are used to quantify such events based on extrapolated observations, or simplifying assumptions are made about extreme precipitation and resulting flood peaks. Continuous simulation (CS) is an alternative approach that better links flood estimation with physical processes and avoids assumptions about antecedent conditions. However, long-term CS has hardly been implemented to estimate rare floods (i.e. return periods considerably larger than 100 years) at multiple sites in a large river basin to date. Here we explore the feasibility and reliability of the CS approach for 19 sites in the Aare River basin in Switzerland (area: 17 700 km2) with exceedingly long simulations in a hydrometeorological model chain. The chain starts with a multi-site stochastic weather generator used to generate 30 realizations of hourly precipitation and temperature scenarios of 10 000 years each. These realizations were then run through a bucket-type hydrological model for 80 sub-catchments and finally routed downstream with a simplified representation of main river channels, major lakes and relevant floodplains in a hydrologic routing system. Comprehensive evaluation over different temporal and spatial scales showed that the main features of the meteorological and hydrological observations are well represented and that meaningful information on low-probability floods can be inferred. Although uncertainties are still considerable, the explicit consideration of important processes of flood generation and routing (snow accumulation, snowmelt, soil moisture storage, bank overflow, lake and floodplain retention) is a substantial advantage. The approach allows for comprehensively exploring possible but unobserved spatial and temporal patterns of hydrometeorological behaviour. This is of particular value in a large river basin where the complex interaction of flows from individual tributaries and lake regulations are typically not well represented in the streamflow observations. The framework is also suitable for estimating more frequent floods, as often required in engineering and hazard mapping.

P09-09 Hybrid work and temporal patterns of sedentary behavior in a 24-hour compositional perspective

BackgroundDuring the COVID-19 pandemic, many white-collar workers were requested to exclusively work from home (WFH), which may have affected their sedentary behavior. In Sweden, having less severe restrictions than many other countries, workers were allowed to alternate between WFH and work at the office (WAO), so called hybrid work. Understanding how hybrid work influences total sedentary behavior and its temporal distribution is an important issue for future health promotion. This study aimed to investigate to what extent office workers changed their temporal pattern of sedentary behavior during days WFH compared to WAO, considering age and gender as potential moderators.MethodsData were collected from May to December 2020 in office workers (n = 199). Their mean age was 42 (SD 10) years and 55% were women. Physical behaviors were measured using a thigh-worn accelerometer (AxivityAX3) for seven consecutive days. A diary identified working hours, time-in-bed and days WFH or WAO. Time-use was classified as short (0-5 min), moderate (5-30 min) and long bouts (>30 min) of sedentary behavior (SB), non-SB, and time-in-bed during workdays (WAO and WFH) and non-workdays. We used Compositional data analysis to express data as 24-hour compositions and linear mixed models to estimate difference in 24-hour compositions between day types (within worker), including age and gender as covariates and moderators.ResultsWe found that workdays (WFH and WAO) were associated with proportionally less time-in-bed relative to time awake, more time SB relative to non-SB, and more time in longer relative to shorter sedentary bouts, compared to non-workdays (all p > 0.001). WFH was associated with more time-in-bed relative to awake and more SB relative to non-SB than WAO (p > 0.05), but the differences for sedentary bouts were not significant. Younger workers and women had more SB, and women accumulated more time than men in longer relative to shorter bouts of SB. However, age and gender did not affect differences between day types.ConclusionsWorking from home influenced 24-hour time-use in office workers by increasing sedentary behavior in total, while its temporal pattern was unchanged. Results contribute to evidence that can support organizational policies on hybrid work.