Scholarlink

Background <p>Medical education is transitioning from traditional didactic teaching to an integrated, technology-enhanced, self-directed model. Studies report high student satisfaction with digital resources, especially YouTube and artificial intelligence–based large language models (LLMs). However, limited comparative evidence exists on resource utilization patterns and their associations with academic performance. This study aims to identify the resources preferred by medical students, assess their satisfaction, and correlate usage patterns with academic performance in Pakistan.</p> Methods <p>This cross-sectional study recruited 380 undergraduate medical students across Pakistan via convenience sampling. The data included demographics, usage frequency, and satisfaction levels for textbooks, institutional lectures, YouTube, LLMs, and medical websites. The primary outcome was the self-reported score in the most recent annual professional examination. Data were analyzed via the Mann–Whitney U test, the Kruskal–Wallis H test, and hierarchical multiple linear regression, controlling for sex, institution type, and year of study.</p> Results <p>Textbooks (55.8%) and YouTube (27.4%) were the most preferred resources. The students rated institutional lectures significantly lower in terms of satisfaction (mean = 2.85) than did YouTube (mean = 4.14) and textbooks (mean = 4.02) (<i>p</i> &lt; 0.001). The regression showed that “sometimes” using YouTube was an independent positive predictor of performance (β = 12.04, <i>p</i> = 0.045). Conversely, higher satisfaction with LLMs was correlated with lower performance; “neutral” to “very satisfied” users scored significantly lower than dissatisfied users did (β ≈–4.90, <i>p</i> &lt; 0.05) though reverse causation was not excluded. Public-sector affiliation was the strongest positive correlate (β = 4.51, <i>p</i> &lt; 0.001).</p> Conclusion <p>Balanced use of YouTube and textbooks is associated with better performance, whereas high satisfaction with LLMs is associated with poorer performance. Institutes should promote structured, multimodal learning to maximize academic potential and future studies should be designed to study these patterns in detail and establish any potential causality.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

YouTube, Large Language Models (LLMs), or Textbooks? Usage Trends and Associations with Academic Performance and Learning Satisfaction among Medical Students in Pakistan: A Cross-Sectional Study

  • 期刊论文

  • 发表日期：

    2026年4月15日

查看全文
  • Muhammad Ahmed,
  • Muhammad Usman,
  • Insha Shafique,
  • Amna Sabir,
  • Ikram Ullah

摘要

Background

Medical education is transitioning from traditional didactic teaching to an integrated, technology-enhanced, self-directed model. Studies report high student satisfaction with digital resources, especially YouTube and artificial intelligence–based large language models (LLMs). However, limited comparative evidence exists on resource utilization patterns and their associations with academic performance. This study aims to identify the resources preferred by medical students, assess their satisfaction, and correlate usage patterns with academic performance in Pakistan.

Methods

This cross-sectional study recruited 380 undergraduate medical students across Pakistan via convenience sampling. The data included demographics, usage frequency, and satisfaction levels for textbooks, institutional lectures, YouTube, LLMs, and medical websites. The primary outcome was the self-reported score in the most recent annual professional examination. Data were analyzed via the Mann–Whitney U test, the Kruskal–Wallis H test, and hierarchical multiple linear regression, controlling for sex, institution type, and year of study.

Results

Textbooks (55.8%) and YouTube (27.4%) were the most preferred resources. The students rated institutional lectures significantly lower in terms of satisfaction (mean = 2.85) than did YouTube (mean = 4.14) and textbooks (mean = 4.02) (p < 0.001). The regression showed that “sometimes” using YouTube was an independent positive predictor of performance (β = 12.04, p = 0.045). Conversely, higher satisfaction with LLMs was correlated with lower performance; “neutral” to “very satisfied” users scored significantly lower than dissatisfied users did (β ≈–4.90, p < 0.05) though reverse causation was not excluded. Public-sector affiliation was the strongest positive correlate (β = 4.51, p < 0.001).

Conclusion

Balanced use of YouTube and textbooks is associated with better performance, whereas high satisfaction with LLMs is associated with poorer performance. Institutes should promote structured, multimodal learning to maximize academic potential and future studies should be designed to study these patterns in detail and establish any potential causality.