Short-term Users Research Articles

Overall Survival and Treatment Patterns Among Patients With Warm Autoimmune Hemolytic Anemia in Sweden: A Nationwide Population-based Study.

Warm autoimmune hemolytic anemia (wAIHA) is a rare autoantibody-mediated disorder, and first-line treatment primarily relies on corticosteroids. This study assessed overall survival (OS) and treatment patterns of wAIHA in Sweden. Adults with ≥ 1 primary diagnosis code for wAIHA (or AIHA plus oral corticosteroids (OCS)/immunosuppressants as sensitivity analyses) between 2011 and 2022 were identified from five Swedish national registers and linked through each patient's unique identity number. Kaplan-Meier curves with log-rank tests and Cox regressions were performed to assess OS for patients with primary versus secondary wAIHA and patients with wAIHA and long-term versus short-term (≥ 3 vs. < 3 months) OCS users. The main analysis included 292 patients; 1791 patients were included in the sensitivity analysis. At a median 3.7-year follow-up, a median OS in primary wAIHA was not reached versus 6.0 years for secondary wAIHA (log-rank test: p = 0.003). Subgroup analyses showed no significant difference in risk of death between long-term and short-term OCS users; however, in the sensitivity analysis, long-term OCS users showed significantly higher risk of death (adjusted hazard ratio: 1.45; 95% confidence interval: 1.180, 1.781; p < 0.001) versus short-term OCS users. Secondary wAIHA or long-term OCS use was associated with lower OS, underscoring the disease burden and unmet need for efficacious wAIHA treatments.

Short term sodium glucose transport protein 2 inhibitors are associated with post contrast acute kidney injury in patients with diabetes

Although sodium–glucose transport protein-2 (SGLT2) inhibitors (SGLT2i) do not increase the risk of acute kidney injury (AKI) in general, they may pose a risk in patients undergoing angiography. This prospective cohort study aimed to evaluate the safety and efficacy of SGLT2i for post-contrast AKI (PC-AKI) in patients with type 2 diabetes mellitus (T2DM). Following screening, 306 patients with T2DM selected to undergo coronary arterial angiography with or without percutaneous intervention were enrolled. Patients were divided into the SGLT2i exposure and non-exposure groups. The primary outcome was PC-AKI, defined as an increase in serum creatinine levels > 0.5 mg/dL (44.2 µmol/L), or 25% above the baseline, within 48–72 h after exposure to contrast medium. The incidence of PC-AKI in the overall T2DM population was 5.2% (16/306). Following 1:1 propensity score matching, the incidence of PC-AKI was significantly higher in the SGLT2i group than in the non-SGLT2i group (10.7% vs. 2.9%; P = 0.027), with an odds ratio of 4.5 (95% confidence interval: 1.0–20.2; P = 0.047). Furthermore, PC-AKI occurred at a higher rate among short-term users of SGLT2i than long-term users (20.5% vs. 3.4%, P = 0.018). Thus, our findings suggest an increased risk of PC-AKI associated with short-term SGLT2i therapy in patients with T2DM.

Investigating Older Adults' Use of a Socially Assistive Robot via Time Series Clustering and User Profiling: Descriptive Analysis Study.

The aging population and the shortage of geriatric care workers are major global concerns. Socially assistive robots (SARs) have the potential to address these issues, but developing SARs for various types of users is still in its infancy. This study aims to examine the characteristics and use patterns of SARs. This study analyzed log data from 64 older adults who used a SAR called Hyodol for 60 days to understand use patterns and their relationship with user characteristics. Data on user interactions, robot-assisted content use, demographics, physical and mental health, and lifestyle were collected. Time series clustering was used to group users based on use patterns, followed by profiling analysis to relate these patterns to user characteristics. Overall, 4 time series clusters were created based on use patterns: helpers, friends, short-term users, and long-term users. Time series and profiling analyses revealed distinct patterns for each group. We found that older adults use SARs differently based on factors beyond demographics and health. This study demonstrates a data-driven approach to understanding user needs, and the findings can help tailor SAR interventions for specific user groups. This study extends our understanding of the factors associated with the long-term use of SARs for geriatric care and makes methodological contributions.

Learning multi-behavior user intent for session-based recommendation

Session-based recommendation makes a recommendation by exploiting short-term user interaction and has become a hot research topic. A quantity of methods have been proposed for session-based recommendation. Although effective, these methods focus on a particular behavior or fail to model the complex relationship between items with multiple behaviors. Thus they can only provide limited recommendation performance. To solve the above problem, we design a Multi-behavior User Intent Recommendation model (MUIR) to produce a recommendation to explore the complicated item–item dependencies with multiple behavior to make more effective recommendation. Technically, MUIR captures item–item cross-session global dependencies and generates behavior-specific user intent representations through a behavior-aware global attention encoder. MUIR employs a low-rank self-attention network to model in-session local dependencies within a session. Moreover, a proposed user behavior discrimination task is auxiliary optimized to disentangle the user intent with multiple behaviors. The extensive experiments show that our MUIR consistently outperforms the various state-of-the-art baselines.

Long-term lorazepam use may be associated with worse long-term outcomes among patients with pancreatic adenocarcinoma

BackgroundLorazepam recently has been reported to alter the tumor microenvironment of pancreatic adenocarcinoma in a murine model. We sought to evaluate whether the use of lorazepam was associated with worse outcomes among patients with pancreatic adenocarcinoma. MethodsMedicare beneficiaries diagnosed with stage I-IV pancreatic adenocarcinoma between 2013 and 2019 were identified from the Surveillance, Epidemiology and End Results-Medicare database. The association of lorazepam prescription relative to overall survival and recurrence-free survival was examined. ResultsAmong 2,810 patients with stage I–III and 10,181 patients with stage IV pancreatic adenocarcinoma, a total of 133 (4.7%) and 444 individuals (4.4%) had a lorazepam prescription before disease diagnosis, respectively. Although the overall lorazepam group had comparable 5-year overall survival (15.0% vs 14.2%, P = .20) and recurrence-free survival (12.7% vs 10.9%, P = .42) with the no-lorazepam group after pancreatic adenocarcinoma resection, individuals with long-term lorazepam prescription (>30 days) had worse 5-year overall survival (9.0% vs 21.0%, P = .02) and recurrence-free survival (6.4% vs 17.1%, P = .009) compared with short-term lorazepam users (≤30 days). Similarly, among patients with metastatic pancreatic adenocarcinoma, individuals with a long-term lorazepam prescription had worse 1-year overall survival (9.7% vs 15.9%, P = .02) compared with patients who had short-term lorazepam prescriptions. On multivariable analysis, long-term lorazepam prescription was independently associated with overall survival among patients with resectable (hazard ratio, 1.82; 95% confidence interval, 1.22–2.74) and metastatic pancreatic adenocarcinoma (hazard ratio, 1.24; 95% confidence interval, 1.02–1.51). ConclusionLong-term lorazepam prescription was associated with worse long-term outcomes among patients who underwent resection for pancreatic adenocarcinoma and patients with metastatic pancreatic adenocarcinoma. These data support the need for further large scale studies to confirm a potential harmful effect of lorazepam among patients with pancreatic adenocarcinoma.

Keywords-enhanced Contrastive Learning Model for travel recommendation

Travel recommendation aims to infer travel intentions of users by analyzing their historical behaviors on Online Travel Agencies (OTAs). However, crucial keywords in clicked travel product titles, such as destination and itinerary duration, indicating tourists’ intentions, are often overlooked. Additionally, most previous studies only consider stable long-term user interests or temporary short-term user preferences, making the recommendation performance unreliable. To mitigate these constraints, this paper proposes a novel Keywords-enhanced Contrastive Learning Model (KCLM). KCLM simultaneously implements personalized travel recommendation and keywords generation tasks, integrating long-term and short-term user preferences within both tasks. Furthermore, we design two kinds of contrastive learning tasks for better user and travel product representation learning. The preference contrastive learning aims to bridge the gap between long-term and short-term user preferences. The multi-view contrastive learning focuses on modeling the coarse-grained commonality between clicked products and their keywords. Extensive experiments are conducted on two tourism datasets and a large-scale e-commerce dataset. The experimental results demonstrate that KCLM achieves substantial gains in both metrics compared to the best-performing baseline methods. Specifically, HR@20 improved by 5.79%–14.13%, MRR@20 improved by 6.57%–18.50%. Furthermore, to have an intuitive understanding of the keyword generation by the KCLM model, we provide a case study for several randomized examples.

Short-term POI recommendation with personalized time-weighted latent ranking

In this paper, we formulate a novel Point-of-interest (POI) recommendation task to recommend a set of new POIs for visit in a short period following recent check-ins, named short-term POI recommendation. It differs from previously studied tasks and poses new challenges, such as modeling high-order POI transitions in a short period. We present PTWLR, a personalized time-weighted latent ranking model that jointly learns short-term POI transitions and user preferences with our proposed temporal weighting scheme to capture the temporal context of transitions. We extend our model to accommodate the transition dependencies on multiple recent check-ins. In experiments on real-world datasets, our model consistently outperforms seven widely used methods by significant margins in various contexts, demonstrating its effectiveness on our task. Further analysis shows that all proposed components contribute to performance improvement.

Novel behavior-enhanced long- and short-term interest model for sequential recommendation

In the realm of modern recommender systems, user-item interaction data often exhibit sequential patterns in relation to various behaviors, such as clicks and purchases on e-commerce platforms. The objective of heterogeneous sequential recommendation (HSR) is to predict each user's next item of interest under a specific behavior based on these interactions. However, existing approaches to HSR struggle to fully capture item transition relationships, as they consider these relationships from a single dimension and at a coarse-grained level. To bridge this gap, we propose the novel Behavior-enhanced Long- and Short-term Interest (BLSI) model, which explores fine-grained item transition relationships in both local and global dimensionalities. At its core, BLSI incorporates a behavior-enhanced self-attention network (BSAN) to capture short-term user preferences. BSAN distinguishes the effects of different behaviors and considers cross-type behavior influences during the linear projection and attention score calculation stages. Additionally, BLSI employs a heterogeneous graph neural network (HGNN) to model long-term user interests by discriminatively aggregating the information of neighboring nodes according to their behavior transition relationships. Furthermore, a gating mechanism is implemented to adaptively fuse short- and long-term preferences for personalized recommendations. Extensive experimental results on three datasets demonstrate that BLSI significantly outperforms state-of-the-art recommendation methods, highlighting the advantages of leveraging sequentiality and behavioral heterogeneity.

The Impact of Opioid Prescription on the Occurrence and Outcome of Pneumonia: A Nationwide Cohort Study in South Korea.

Opioids are known to cause respiratory depression, aspiration, and to suppress the immune system. This study aimed to investigate the relationship between short- and long-term opioid use and the occurrence and clinical outcomes of pneumonia in South Korea. The data for this population-based retrospective cohort analysis were obtained from the South Korean National Health Insurance Service. The opioid user group consisted of those prescribed opioids in 2016, while the non-user group, who did not receive opioid prescriptions that year, was selected using a 1:1 stratified random sampling method. The opioid users were categorized into short-term (1-89 d) and long-term (≥90 d) users. The primary end point was pneumonia incidence from January 1, 2017-December 31, 2021, with secondary end points including pneumonia-related hospitalizations and mortality rates during the study period. In total, 4,556,606 adults were enrolled (opioid group, 2,070,039). Opioid users had a 3% higher risk of pneumonia and an 11% higher risk of pneumonia requiring hospitalization compared to non-users. Short-term users had a 3% higher risk of pneumonia, and long-term users had a 4% higher risk compared to non-users (P < .001). Additionally, short-term users had an 8% higher risk of hospital-treated pneumonia, and long-term users had a 17% higher risk compared to non-users (P < .001). Both short- and long-term opioid prescriptions were associated with higher incidences of pneumonia and hospital-treated pneumonia. In addition, long-term opioid prescriptions were linked to higher mortality rates due to pneumonia.

The association between proton-pump inhibitor use and recurrence of hepatocellular carcinoma after hepatectomy.

The association between long-term proton-pump inhibitors (PPIs) use and malignancies had long been discussed, but it still lacks consensus. Our study investigated the association between PPI use and hepatocellular carcinoma (HCC) recurrence following curative surgery. We retrospectively enrolled 6037 patients with HCC who underwent hepatectomy. Patients were divided into four groups according to their PPI usage. (non-users: <28 cumulative defined daily dose [cDDD]; short-term users: 28-89 cDDD; mid-term users: 90-179 cDDD, and long-term users: ≥180 cDDD, respectively). Recurrence-free survival (RFS) and overall survival (OS) were analyzed using Kaplan-Meier method and Cox proportional hazard models. Among the 6037 HCC patients, 2043 (33.84%) were PPI users. PPI users demonstrated better median RFS (3.10years, interquartile range [IQR] 1.49-5.01) compared with non-users (2.73years, IQR 1.20-4.74; with an adjusted hazard ratio [aHR] of 0.57, 95% confidence interval [CI] 0.44-0.74, P<0.001). When considering the cumulative dosage of PPI, only long-term PPI users had significant lower risk of HCC recurrence than non-PPI group (adj-HR: 0.50; 95% CI: 0.35-0.70; P<0.001). Moreover, the impact of long-term PPIs use on improving RFS was significant in most of the subgroup analysis, except in patients with advanced tumor stages, with non-cirrhosis, or with a history of chronic kidney disease. However, there were no significant differences in median OS between PPI users and non-users (4.23years, IQR 2.73-5.86 vs 4.04years, IQR 2.51-5.82, P=0.369). Long-term PPI use (≥180 cDDD) may be associated with a better RFS in HCC patients after hepatectomy.

Unmasking user vulnerability: investigating the barriers to overcoming dark patterns in e-commerce using TISM and MICMAC analysis

PurposeE-commerce companies use dark patterns to manipulate customer decisions to survive in the crowded online market and make profit. Although some online customers are aware of the dark patterns, they cannot overcome such manipulations. Therefore, the purpose of this study is to identify and model the barriers to overcoming dark patterns using total interpretive structural modeling (TISM).Design/methodology/approachBarriers to overcoming dark patterns were identified from the extant literature and were validated by a panel of 18 domain experts. In the modeling phase, TISM technique was used to identify the relationships between the barriers and assign priority to the barriers. Finally, the barriers were plotted and classified into three categories.FindingsUser unawareness, trust in brands and normalization of aggressive marketing were found to be the highest priority barriers. Whereas, designer bias, user fatigue, short-term user benefits and design complexity were identified as the most challenging barriers because they have least dependence over the other barriers.Research limitations/implicationsBecause TISM results are based on the opinion of domain experts, other statistical techniques could be applied for validation.Practical implicationsThis study would educate online customers, while assisting online user communities and regulatory bodies to devise strategies to overcome dark patterns. Additionally, business managers could use the study’s findings to encourage designers to embrace ethical design methods as a competitive advantage.Originality/valueThis study contributes to the research as it is first of its kind to examine the link between dark pattern barriers.

Safety education at research facilities with radiation sources and short-term facility users – Current design and practice and possibilities for improvement

Research facilities such as spallation sources and synchrotrons generate radiation for use in atomic-level or molecular-scale experiments. These facilities can be viewed as complex safety-critical systems. An important aspect of the safety management of such systems is the short safety education and training programme the users are required to undergo in order to gain facility access. As research on the topic is limited, this study aimed to increase the knowledge about current education design and practice using the perspectives of safety science and pedagogy. Study objectives were to identify preconditions that impact the safety education design, to describe current design and practice of the safety education, and to identify weaknesses and possibilities for improvement. Site visits with a total of 20 interviews were performed at three research facilities. The results show the need for sufficient resources to maintain learning activities for users, provide pedagogical continuing education for educators, and maintain safety culture-enhancing activities to meet the challenges of having large numbers of short-term facility users. Increased focus should be placed on safety-related competence needs and the mapping of these to match the competence of individual users. New thinking and innovation can benefit the design and provision of such education activities, based on both socio-technical system and system safety perspectives.

Prescription Patterns and Predisposing Factors of Benzodiazepine and Z-Hypnotic Use During Pregnancy: A Nationwide Cohort Study.

The use of benzodiazepines and Z-hypnotics during pregnancy has raised significant concerns in recent years. However, there are limited data that capture the prescription patterns and predisposing factors in use of these drugs, particularly among women who have been long-term users of benzodiazepines and Z-hypnotics before pregnancy. This population-based cohort study comprised 2 930 988 pregnancies between 2004 and 2018 in Taiwan. Women who were dispensed benzodiazepines or Z-hypnotics during pregnancy were identified and further stratified into groups based on their status before pregnancy: long-term users (with a supply of more than 180 days within a year), short-term users (with a supply of less than 180 days within a year), and nonusers. Trends in the use of benzodiazepines or Z-hypnotics and concomitant use with antidepressants or opioids were assessed. Logistic regression models were utilized to identify factors associated with use of these drugs during pregnancy, and interrupted time series analyses (ITSA) were employed to evaluate utilization patterns of these drugs across different pregnancy-related periods. The overall prevalence of benzodiazepine and Z-hypnotic use was 3.5% during pregnancy. Among prepregnancy long-term users, an upward trend was observed. The concomitant use of antidepressants or opioids among exposed women increased threefold (from 8.6% to 23.1%) and sixfold (from 0.3% to 1.7%) from 2004 to 2018, respectively. Women with unhealthy lifestyle behaviors, such as alcohol abuse (OR 2.48; 95% CI, 2.02-3.03), drug abuse (OR 10.34; 95% CI, 8.46-12.64), and tobacco use (OR 2.19; 95% CI, 1.96-2.45), as well as those with psychiatric disorders like anxiety (OR 6.99; 95% CI, 6.77-7.22), insomnia (OR 15.99; 95% CI, 15.55-16.45), depression (OR 9.43; 95% CI, 9.07-9.80), and schizophrenia (OR 21.08; 95% CI, 18.76-23.69), and higher healthcare utilization, were more likely to use benzodiazepines or Z-hypnotics during pregnancy. ITSA revealed a sudden decrease in use of benzodiazepines and Z-hypnotics after recognition of pregnancy (level change -0.55 percentage point; 95% CI, -0.59 to -0.51). In contrast, exposures to benzodiazepines and Z-hypnoticsincreased significantly after delivery (level change 0.12 percentage point; 95% CI, 0.09 to 0.16). In this cohort study, an increased trend of benzodiazepine and Z-hypnotic use during pregnancy among prepregnancy long-term users, as well as concomitant use with antidepressants or opioids were found. The findings have highlighted the existence of various risk factors associated with the use of these drugs during pregnancy. Utilization patterns varied across different stages of pregnancy, highlighting the need for prescription guidelines and educational services for women using these drugs during pregnancy.

Finding Advisors in Collaborative Environment with Fine-Grained Knowledge Sharing

Abstract: As the digital world continues to expand, traditional search engines face difficulties in the ability to provide users with excellent and rapid information retrieval. Additionally, users have shown a low tolerance for retrieving irrelevant information and are increasingly expecting to be served with personalized information that meets their needs and preference. Personalized environments are of many advantages to applications such as e-commerce and research due to their need to provide the best possible results to the users as fast as possible. Personalized web environments that customize search results to specific users based on their behavior and preference while searching are becoming increasingly popular to solve this problem. Short-term and long-term user preferences are constructed from data on actions and top queries, visited content, and user preferences. Lots of websites already use this kind of personalized search, and it's making a big difference in how people use the internet

A Long-and Short-term Hot User Identification Method Based on Local Outlier Density

Power systems have many users; however, there are relatively few users focused on high-quality power supplies (namely, hot users), such as external query users, electricity theft, and information fraud from billions of users. Starting from power load data, this paper postulates that abnormal electricity load users are essential components of hot users. According to the characteristics of regional power loads, this paper first builds a cross-judgment method for abnormal loads to improve the identification accuracy, including spatial and temporal abnormal load detection methods. As the spatial method, the abnormal load detection based on the local outlier factor (LOF) algorithm detects different users from similar users using several constructed dimensional features. As the temporal method, dynamic abnormal load detection based on the short-term sequential (SS) algorithm is designed to improve the LOF missing alarm rate and computing efficiency with multi-dimensional features. This also accomplished abnormal load detection from historical dimensions. A long- and short-term hot user detection method is designed based on the outlier density in response to the insufficient support of single-point abnormal loads to judge hot users. Finally, experiments on the user load data in a certain region within a year test the abnormal detection method for long- and short-term hot user detection based on local outlier density (LOD). The experimental results indicate that this method can detect abnormal loads and search hot users. The long- and short-term hot user detection method based on the LOD can meet practical needs when only using user load data and has good generalization abilities.

Carbon emissions of power transmission and transformation projects in the whole life cycle for smart sustainable energy systems

The study investigates the optimization of life cycle carbon emissions in smart sustainable energy systems through power transformation and transmission project power load predictions. Firstly, a multi-task learning-based short-term user load forecasting technique is developed, where the power load curves of multiple residential customers are grouped and classified using the K-means clustering method. Additionally, the Bidirectional Long Short-Term Memory (BiLSTM) technique is introduced to anticipate the power load intelligently. Secondly, a life cycle carbon emission assessment model for the power transmission and transformation project (PTTP) is constructed based on the life cycle assessment (LCA) method, which divides the project's life cycle into four stages: production, installation and construction, operation and maintenance, and demolition. Finally, an experimental evaluation of this model is conducted. The results demonstrate that compared with the baseline model Long Short-Term Memory (LSTM), this model achieves a significantly lower average Mean Absolute Error (MAE) at 3.62% while achieving significantly higher accuracy in power load forecasting at 94.34%. A comprehensive examination of carbon emissions across all four phases reveals that overall carbon emissions are highest during the operation and maintenance stage followed by the equipment production stage and installation/construction stage, with the lowest overall carbon emissions observed. Hence, this study endeavors to forecast power load demand with precision and identify the principal determinants of carbon emissions in power engineering. By discerning and managing these key factors, an optimal, energy-efficient intelligent power load scheme can be derived.

The Effects of Clopidogrel Duration on Carotid Artery In-Stent Restenosis

ObjectiveThere is limited data supporting a specific duration for dual antiplatelet therapy in carotid artery stenting (CAS), and most clinical evidence is derived from studies involving coronary interventions. As a result, the appropriate duration of dual antiplatelet therapy after CAS has yet to be determined. We aimed to elucidate whether the duration of dual antiplatelet therapy played a role in the rate of carotid in-stent restenosis. MethodsA retrospective analysis of all patients who underwent CAS at our institution over a 20-year period (1996-2016) was performed (n=279). Patients who did not complete their follow-up duplex studies or were not discharged on clopidogrel were excluded from the study. Patients were separated into short-term (<6 weeks, n=159) and long-term (>6 weeks, n=112) clopidogrel users based on duration of therapy. We defined clinically significant in-stent restenosis as >50% restenosis (PSV = 224 cm/s) in symptomatic patients and >80% restenosis (PSV = 325 cm/s) in asymptomatic patients status-post prior CAS based on published velocity criteria. Rates of in-stent restenosis at 1-year, 2-year, and 5-year intervals were analyzed between the two groups using chi-squared analysis. ResultsDemographic information was largely similar between the two groups; however, short-term clopidogrel users were more likely to have a history of atrial fibrillation (9.43% vs. 1.68%, p=0.008) and were less likely to have a history of CABG (16.35% vs. 29.41%, p=0.009), diabetes (33.34% vs. 49.58%, p=0.006) and CAD (50.31% vs. 63.03%, p=0.035). All patients were on long-term aspirin therapy. There was no significant difference between overall rates of in-stent restenosis between the short-term and long-term clopidogrel users (5.03% vs. 9.24%, p=0.168) within 5 years of the index procedure. Similar results were observed when these groups were evaluated at 1-year (5.61 % vs. 3%, p=0.321), 2-year (2.02% vs. 6.59%, p=0.072), and 5-year (2.24% vs. 3.57%, p=0.635) follow-up. ConclusionNo statistically significant difference was observed in the rate of in-stent restenosis after CAS between short-term and long-term clopidogrel therapy. Patients in whom there is no other indication for longer duration clopidogrel therapy may be considered for shorter duration course of dual antiplatelet therapy following CAS.

SR-MACL: Session-Based Recommendation with Multi-layer Aggregation Augmentation in Contrastive Learning

Session-based recommendation (SBR) aims to predict the next item of interest in chronological order based on a given sequence of short-term behaviour of anonymous users. Due to the limited data available for short-term user interactions, its performance is more susceptible to data sparsity problems than traditional recommendation methods. Contrastive learning is often used to solve the data sparsity problem due to its ability to extract general features from the raw data. Existing session-based recommendation methods based on graph contrastive learning typically build graph contrastive learning by using information from other sessions to generate augmented views. While this avoids the problem that the use of dropout in traditional contrast learning methods can cause damage to the session context, it inevitably introduces irrelevant item information, which interferes with accurately modelling user interests and leads to sub-optimal model performance. To address these issues, we propose a new session recommendation method based on multi-layer aggregation augmentation contrastive learning, namely SR-MACL. In SR-MACL we construct a contrastive view by adding noise to the embedding representation and forming a contrastive embedding representation by multi-layer aggregation, which not only effectively solves the problem that traditional graph enhancement methods can destroy the context of the whole session, but also avoids the interference of irrelevant items. Experimental results on three real datasets have shown that SR-MACL can improve the accuracy of recommendation results and predict the user's next interaction more effectively.

Aggregating knowledge and collaborative information for sequential recommendation

Sequential recommendation aims to predict users’ future activities based on their historical interaction sequences. Various neural network architectures, such as Recurrent Neural Networks (RNN), Graph Neural Networks (GNN), and self-attention mechanisms, have been employed in the tasks, exploring multiple aspects of user preferences, including general interests, short-term interests, long-term interests, and item co-occurrence patterns. Despite achieving good performance, there are still limitations in capturing complex user preferences. Specifically, the current structures of RNN, GNN, etc., only capture item-level transition relations while neglecting attribute-level transition relations. Additionally, the explicit item relations are studied using item co-occurrence modules, but they cannot capture the implicit item-item relations. To address these issues, we propose a knowledge-augmented Gated Recurrent Unit (GRU) to improve the short-term user interest module and adopt a collaborative item aggregation method to enhance the item co-occurrence module. Additionally, our long-term interest module utilizes a bitwise gating mechanism to select historical item features significant to users’ current preferences. We extensively evaluate our model on three real-world datasets alongside competitive methods, demonstrating its effectiveness in top K sequential recommendation.

Trends of anti-seizure medication prescribing pattern in traumatic brain injury patients for the prevention of posttraumatic seizure in Taiwan

Traumatic brain injury (TBI) patients are recommended to receive anti-seizure medication (ASM) as posttraumatic seizure (PTS) prophylaxis. However, the utilization of ASM, including the prescription patterns and associated clinical characteristics, is limited in Taiwan. Thus, this study aimed to investigate the ASM trends and clinical characteristics. This retrospective cohort study enrolled TBI patients who received levetiracetam, phenytoin, and valproic acid during hospitalization using the National Health Insurance Research Database between 2012 and 2019. The primary outcome was the trend of the ASMs based on the index year. The duration of levetiracetam prescription was categorized as short-term (seven days or less) or long-term (more than seven days). Logistic regression identified the factors associated with long-term usage. A total of 64,461 TBI patients were included. Levetiracetam usage increased yearly, while phenytoin declined. Among the levetiracetam users, 5681 (30.38%) were short-term users, and 13,016 (69.62%) were long-term users. Diagnoses of contusions, intracranial hemorrhage, other intracranial injuries, receiving operations, and a history of cerebrovascular disease were significantly associated with longer duration. Conclusions This study revealed the rising trend of levetiracetam usage, indicating its potential as an alternative to phenytoin. TBI patients with more severe conditions were more likely to receive longer prescriptions.