How Do Sleep and Gender Impact Recognition of Emotions and Faces?

Overview

The purpose of this quasi -experimental study was to determine the impact of sleep and gender on emotional and facial recognition. The hypotheses of this study were that inadequate sleep would have a negative impact on facial and emotional recognition and t hat women would outperform men on facial and emotional recognition tasks, regardless of sleep adequacy. There were 48 participants in this study, with twenty -six females and twenty -two males. The average participant was sixteen years old. All participants were from Roanoke Valley Governor’s School and gave informed consent before participating in this study. First, participants completed a demographic questionnaire and questions from the Pittsburgh Sleep Quality Index (PSQI). Then, participants completed th e Emotion Recognition Task (ERT) and the Glasgow Face Matching Task 2 (GFMT2). Results suggest females and males had similar sleep quality, with no significant difference between the mean PSQI scores (p = 0.444). Female GFMT2 scores were significantly different than male GFMT2 scores (p = 0.013), with females having a mean of 86.54 and males having a mean of 82.39. Additionally, female and male scores on the ERT were significantly different (p = 0.014), with females having a mean score of 64.58 and males having a mean score of 55.7. Sleep did not have a significant impact on facial or emotional recognition (p = 0.261 and 0.645, respectively). Overall, sleep did not influence facial or emotional recognition, but gender did; females outperformed males on both recognition tasks. Key Drivers of Patient Loyalty Hailey Kim Marriotts Ridge High School, Marriottsville, MD Patient loyalty is a critical component of healthcare quality, influencing patient adherence, engagement, and continuity of care. While doctor reputation has been recognized as an essential factor in healthcare decision -making, its direct impact on patient loyalty remains unclear. This study examines the relationship between doctor reputation and patient loyalty, proposing that patient-centered care serves as a mediator in this relationship. A quantitative survey -based approach was employed, recruiting 250 U.S. patients through Amazon Mechanical Turk (MTurk). Participants completed validated 7-point Likert-type scales measuring doctor reputation, patient- centered care, and patient loyalty. Reliability analyses confirmed strong internal consistency for all constructs (Cronbach’s alpha > .80). Hypothesis testing using linear regression supported the positive relationship between doctor reputation and patient loyalty (t = 15.96, p < .001). Mediation analysis using the Hayes (2017) PROCESS macro confirmed the ind irect effect of patient - centered care (indirect effect = .48, 95% CI [.34, .62]), supporting the mediating role of patient - centered care in strengthening the relationship between doctor reputation and patient loyalty. These findings suggest that while doct or reputation establishes an initial foundation for trust, patient-centered care is a key determinant of long -term patient commitment. The results emphasize the importance of integrating patient -centered practices to enhance loyalty beyond reputation alone. Future research should explore longitudinal effects and expand the study across diverse healthcare settings to improve generalizability. Continual Learning-Based Approach to Enhancing Optical Character Recognition for Low-Resource Languages Aiden Ko Korea International School, Seongnam-si, Republic of Korea The state-of-the-art optical character recognition (OCR) tools are often only effective on a narrow scope of languages and scripts, which limits their applicability for many users worldwide. This is largely due to a lack of data used in training the machin e learning models for OCR tasks, giving such languages the title of “low -resource languages.” Even when additional data for such languages become available, traditional methods for refining such tools with new data are not efficient, further reducing the i ncentive for development in practical applications. A class of learning approaches called continual learning offers a remedy through its potential to incorporate new data into existing deep learning algorithms efficiently and accurately without significant time or resource requirements. We explore one continual learning approach, namely the replay method, as a means to improve an OCR engine’s performance for Hangul, the Korean writing system. We experimentally evaluate the effectiveness and efficacy of cont inual learning by measuring the performance of a deep neural network when it is trained through the replay method. Our findings indicate that continual learning, implemented with certain sampling rates of the replay method, demonstrates promising results rates in advancing OCR for Hangul, thereby possessing potential to help resolve the low-resource problem.

Competition history

• JSHS 2025 Category not listed

Resources

• National JSHS abstract book ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.