32584 research outputs found
Sort by
Cluster analysis and concept drift detection in malware
Get PDFConcept drift refers to gradual or sudden changes in the properties of data that affect the accuracy of machine learning models. In this paper, we address the problem of concept drift detection in the malware domain. Specifically, we propose and analyze a lightweight clustering-based approach to detecting concept drift. Using a subset of the KronoDroid dataset, malware samples are partitioned into temporal batches and analyzed using MiniBatch K-Means clustering. The silhouette coefficient is used as a metric to identify points in time where concept drift has likely occurred. To verify our drift detection results, we train learning models under three realistic scenarios, which we refer to as static training, periodic retraining, and drift-aware retraining. In each scenario, we consider four supervised classifiers, namely, multilayer perceptron (MLP), support vector machine (SVM), random forest, and XGBoost. Experimental results demonstrate that drift-aware retraining guided by silhouette coefficient thresholding achieves classification accuracy far superior to static models and, on average, within 0.5% of periodic retraining, while also being far more efficient than periodic retraining. These results provide strong evidence that our clustering-based approach is effective at detecting concept drift, while also illustrating a highly practical and efficient fully automated approach to improved malware classification via concept drift detection
Experiences of U.S. frontline physicians during the COVID-19 pandemic: a qualitative study
Get PDFBackground: The COVID-19 pandemic caused profound and rapid changes in patient care and healthcare system organization. There is a compelling need for insight into the challenges that confronted physicians during the early phase of the pandemic to identify successful adaptations and strategies that minimize disruption to patient care and protect clinician wellbeing. The purpose of this study was to understand physicians’ lived experiences of providing patient care during the early COVID-19 pandemic. Methods: This qualitative, descriptive study used a thematic analysis approach. The sample included 17 physicians from five specialties with direct care experience of COVID-19 patients (infectious disease, primary care, emergency medicine, critical care, and hospitalists). Participants were identified through snowball sampling. Data were collected through focus groups and interviews in May and June 2020 and analyzed with an inductive and deductive approach using thematic analysis. Results: Three overarching themes relating to patient care delivery during the ongoing COVID-19 pandemic were identified: facilitators, barriers, and acute stressors. Facilitator subthemes included: organizational logistical and operational support, organizational support for self-care and wellness, and peer and family support/debriefing. Barrier subthemes included: lack of clear and consistent governmental guidelines and organizational support, uncertainty resulting from poor communication or lack of information, and interpersonal barriers to physician self-care and wellbeing. Stressor subthemes included: concern about exposure, feeling unprepared, and anticipating the worst. Conclusions: Physicians reported that both patient care and their own wellbeing were greatly impacted by organizational and systems level facilitators and barriers. Findings from this study can inform the creation of best practices, tools, and strategies that can assist with future emergency preparedness and pandemic response planning efforts
Using Presence Circles to identify strategies to enhance nurse presence at discharge: a focus group study with nursing students
Get PDFBackground: Nurse communication of patient needs at discharge is critical to ongoing care, but system-level demands often prohibit comprehensive discharge conversations. Caregivers of discharged patients frequently report feeling underprepared to meet patient needs. Meaningful interpersonal encounters, or presence, are known to enhance clinical interactions amidst system-level demands, which could help improve caregiver preparedness and patient care. Purpose: To explore and synthesize examples of nurse presence during discharge conversations through Presence Circles (structured focus groups) to provide recommendations for enhancing high-quality information at discharge. Methods: In a secondary analysis of data from a larger study based on the Nurse Presence Framework, nursing students (N = 14) from a Northern California school were asked to participate in two Nurse Presence Circles. Audio recordings from 10 Presence Circles were transcribed and analysis was conducted according to the five practices of the Nurse Presence framework: Prepare with Intention, Listen Intently and Completely, Agree on What Matters Most, Connect with the Story, and Evolve System-Level Change. Results: Within each of the five nurse presence practices, strategies and challenges were collapsed into broader themes that served as recommendations for enhancing the exchange of high-quality information at discharge. Conclusions: Presence Circles offered a useful space to share strategies and identify system changes that could advance the exchange of high-quality information at discharge. We have provided a synthesis of recommendations for nurses, particularly those new to discharge conversations or early in their nursing career, demonstrating the need to engage nursing students about discharge conversations and introduce considerations related to health care systems and policy to better support the discharge conversation experience
Deep Learning for Non-Invasive Blood Pressure Monitoring: Model Performance and Quantization Trade-Offs
Get PDFThe development of non-invasive blood pressure monitoring systems remains a critical challenge, particularly in resource-constrained settings. This study proposes an efficient deep learning framework integrating Edge Artificial Intelligence for continuous blood pressure estimation using photoplethysmography (PPG) signals. We evaluate three architectures: a residual-enhanced convolutional neural network, a transformer-based model, and an attentive BPNet. Using the MIMIC-IV waveform database, we implement a signal processing pipeline with adaptive filtering, statistical normalization, and peak-to-peak alignment. Experiments assess varying temporal windows (10 s, 20 s, 30 s) to optimize predictive accuracy and computational efficiency. Attentive BPNet achieves the best performance, with systolic blood pressure (SBP) estimation yielding a mean absolute error (MAE) of 6.36 mmHg, diastolic blood pressure (DBP) an MAE of 4.09 mmHg, and mean arterial pressure (MBP) an MAE of 4.56 mmHg. Post-training quantization reduces the model size by 90.71% (to 0.13 MB), enabling deployment on Edge devices. These findings demonstrate the feasibility of deploying deep learning-based continuous blood pressure monitoring on edge devices. The proposed framework provides a scalable and computationally efficient solution, offering real-time, accessible monitoring that could enhance hypertension management and optimize healthcare resource utilization
Unmasking Public Sentiment: A Sample Efficient Approach to Analyzing Twitter Opinion on U.S. Aid to Ukraine
Get PDFThe U.S. aid to Ukraine is a bipartisan topic of extreme socio-political importance. While several organizations have conducted surveys to understand the public stance on this topic, there is no research to date that analyses public opinion on social media, possibly due to the lack of annotated data. Therefore, this research compares several sample-efficient methods (including in-context learning) to analyze tweet sentiments with minimal training data. First, we collect 11,289 tweets about the U.S. aid to Ukraine and mapped them to U.S. states. Next, we explore three different approaches to sentiment analysis: tool-based, embedding-based, and prompt-based. Our results indicate that GPT-4 Few Shot improves accuracy by 121.8% and 77.5% over TextBlob and Vader, respectively. Our geospatial analysis shows that Indiana has the most negative normalized net sentiment (NNS) of -0.83, while Vermont has the most positive NNS of +0.33. Finally, we perform a detailed thematic analysis to identify the common arguments that support or oppose the aid. We highlight that our results do not correlate with media surveys, possibly due to the presence of echo chambers and algorithmic biases
Spartan Daily, September 23, 2025
Get PDFVolume 165, Issue 12https://scholarworks.sjsu.edu/spartan_daily_2025/1055/thumbnail.jp
Developing Culturally-tailored Diabetes Friendly and Heart Healthy Recipes for South Asians: Sensory Evaluation and Implications for Dietary Interventions
Get PDFSouth Asians have a higher risk of developing type 2 diabetes than manyother racial or ethnic groups. Dietary patterns are a key, modifiable risk factor, and cultural influences strongly shape food choices and lifestyle decisions. Effective, long-term behavior change requires culturally meaningful interventions. This study aimed to develop culturally tailored recipes that comply with the American Diabetes Association (ADA) and American Heart Association (AHA) guidelines.
Recipes were selected from the published South Asian Carbohydrate Counting Tool, created by the researcher, and modified to enhance nutrient composition. Adjustments included ingredient substitutions, reduction or elimination of fat, salt, and sugar, and additions to increase fiber content. Changes were introduced gradually, with each version analyzed using FoodLabelMaker software. Fourteen modified recipes were prepared and tested in a lab kitchen. Twelve participants evaluated them based on taste, texture, mouthfeel, flavor, color, odor, portion size, and overall acceptability using a f ive-point Likert scale with a score of 1 as unacceptable and 5 as excellent.
Descriptive statistics were generated, and Spearman correlation coefficients assessed the relationship between overall liking and participants’ intention to adopt recommended dietary behaviors. Results indicated that 10 of the 14 recipes met sensory acceptability criteria (score 3.0 or more) and overall liking was significantly correlated with intentions to improve diet quality.
These findings suggest that consumer acceptance evaluations could enhance interventions aimed at improving diet quality. By increasing acceptance of healthier recipes, such interventions may be more effective in promoting long-term dietary improvements among the South Asian community
Effectiveness and Assessment of a Culturally-tailored Carbohydrate Counting Tool for South Asians with Diabetes: A Practitioner’s Evaluation
Get PDFSouth Asians are at a higher risk for type 2 diabetes compared to many other racial and ethnic groups. Dietary patterns play a significant role and are a modifiable risk factor. This study aimed to evaluate a South Asian carbohydrate counting tool developed by the researcher, assessed by Registered Dietitians Nutritionists and Certified Diabetes Care and Education Specialists. Using nutrition communication theory and the transtheoretical model, the tool included a comprehensive list of commonly consumed South Asian foods and sample vegetarian and non-vegetarian meals (1800e2100 kcal). Sixty-five participants received the tool and a training slide deck. A 40-item questionnaire assessed the dietary knowledge and evaluated the tool’s accuracy, relevance, clarity, cultural appropriateness, and usability. Demographics, knowledge of South Asian diets, perception of adequacy, evaluation of readability, content, and format criteria were measured using 5-point frequency and attitudinal scales, with three open-ended questions capturing user experience. After four weeks, 43 participants completed the assessment. All participants strongly agreed or agreed that the tool and slide deck were useful for client care. The training tool and menus were rated as applicable (97%), relevant (96%), consistent (89%), well-organized (100%), and culturally appropriate (100%). Additionally, 89% of participants considered readability and would recommend the tool to others. Results also indicated that the participants prioritized content accuracy (mean ¼ 4.40) over readability (mean ¼ 3.52) and format (mean ¼ 3.25).
This study validated the effectiveness of the carbohydrate counting tool for South Asians and highlighted its cultural relevance in diabetes meal management and care
Separation and Surface Examination of Spacecraft Cabin Particulates
Get PDFParticulate matter in the atmosphere is a known detriment to human health, but several factors affect any particulate’s particular toxicity including particle size, surface area, and chemical composition. Extensive studies have been conducted to examine particulate hazards on Earth, but in crewed spacecraft, like the International Space Station (ISS), the particulate environment is wholly unique due to both the controlled environment and lack of gravity. Atmospheric sampling studies are underway, but another convenient source of airborne particulates for analysis is the contents of the vacuum bags used by the astronauts to clean the air vents. However, no method currently exists to isolate the respirable portion for further study. The goal of this study is therefore twofold: separate the respirable PM10 and PM2.5 particulates from an ISS vacuum bag and analyze their surface to determine potential toxicity. A repeatable separation method was developed and refined using a simulant of house dust collected from a home vacuum. Respirable particulate yield from separation of simulant samples was significant, but several differences were observed between the simulant and ISS particulates. Specific surface area analysis of the isolated particulates was pursued. Optical microscopy, scanning electron microscopy and energy-dispersive x-ray spectroscopy provided particle morphology and relative elemental composition for both simulant and ISS samples to determine potential toxicity. Several lessons learned and paths forward for improved separation and analysis are discussed
From disciplinary enthusiasm to soulless tasks: norms behind computing educators\u27 emotion display
Get PDFComputing educators\u27 emotion display is regulated by various norms and conventions. These (unwritten) rules affect which emotions educators feel comfortable expressing to different extents when students can perceive such displays. We draw on emotions and norms literature to investigate higher education computing educators\u27 perceptions of what kinds of norms they have as criteria when considering which emotions they are (not) comfortable showing when teaching. Based on the qualitative content analysis of 22 interviews with computing educators from seven countries, we present various–sometimes connected or conflicting–norms that influence emotion display. These norms comprise moral, national, societal, professional, and affiliation norms that are salient in different ways for different educators. The findings contribute to a more nuanced understanding of various combinations of nested, context-dependent, and partly conflicting norms that guide educators\u27 emotion display, an understanding of educators\u27 work that goes beyond cognitive aspects