國立高雄大學圖資館 |

登入

回首頁

Transforming Social Media Data Into Mental Health Insights: Decoding Drug Abuse Patterns With Deep Learning and LLMs /

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Transforming Social Media Data Into Mental Health Insights: Decoding Drug Abuse Patterns With Deep Learning and LLMs /Julina Maharjan.
作者:	Maharjan, Julina,
面頁冊數:	1 electronic resource (202 pages)
附註:	Source: Dissertations Abstracts International, Volume: 87-06, Section: A.
附註:	Advisors: Jin, Ruoming Committee members: Kenne, Deric; Guan, Qiang; Maletic, Jonathan; Gandolfi, Enrico.
Contained By:	Dissertations Abstracts International87-06A.
標題:	Computer science.
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32455547
ISBN:	9798270221188

Transforming Social Media Data Into Mental Health Insights: Decoding Drug Abuse Patterns With Deep Learning and LLMs /
Maharjan, Julina,

Transforming Social Media Data Into Mental Health Insights: Decoding Drug Abuse Patterns With Deep Learning and LLMs /Julina Maharjan. - 1 electronic resource (202 pages)

Source: Dissertations Abstracts International, Volume: 87-06, Section: A.

Online social media discourse data offers unprecedented opportunities to study drug use behaviors and related mental health signals at scale. This dissertation, Transforming Social Media Data into Mental Health Insights: Decoding Drug Abuse Patterns with LLMs, advances methodological frameworks that leverage deep learning, natural language processing (NLP), and large language models (LLMs) to decode complex, drug-related discourse on platforms such as Twitter/X.The first study implements large-scale deep learning pipelines to analyze drug/substance use patterns before, during, and after the COVID-19 pandemic. By processing millions of posts with a state-of-the-art Transformer-based language model (RoBERTa) and iterative Human-in-The-Loop fine-tuning, this study uncovers significant temporal shifts, particularly heightened alcohol discourse during pandemic peaks, and demonstrates how scalable NLP can map population-level health behaviors. The second study develops demographic-aware NLP models, integrating sentiment and emotion classifiers with demographic inference algorithms to capture nuanced differences across age, gender, and race in substance use discussions. These results were validated against national survey baselines, demonstrating how deep learning can generate insights of direct, actionable relevance to public health policy and interventions.Building on these core studies, the dissertation presents two additional methodological contributions. First, it provides a psychometric evaluation of LLM-derived embeddings for personality trait prediction, rigorously validating their reliability and interpretability against established psychological instruments. This work highlights that LLM embeddings sufficiently capture key emotion features, which are otherwise required in traditional personality prediction models. Second, it extends the framework to investigate the intersection of personality and substance use discourse during COVID-19, demonstrating the utility of personality-aware NLP models for behavioral health research.By uniting large-scale social media analysis with cutting-edge NLP and LLM methodologies, this dissertation establishes a robust computational foundation for decoding complex human language and behavior. The contributions demonstrate how methodological innovation; from scalable deep learning pipelines to theoretical advances in model adaptation, can transform raw digital traces into actionable insights for public health policy, targeted interventions, and the next generation of social media? based mental health research.

English

ISBN: 9798270221188Subjects--Topical Terms:

199325
Computer science.
Subjects--Index Terms:

Drug use

Transforming Social Media Data Into Mental Health Insights: Decoding Drug Abuse Patterns With Deep Learning and LLMs /
LDR:04175nmm a22004453i 4500 001 693816
005 20260511063029.5
006 m o d
007 cr|nu||||||||
008 260819s2026 miu||||||m |||||||eng d
020 $a 9798270221188
035 $a (MiAaPQD)AAI32455547
035 $a (MiAaPQD)OhioLINK:kent1765287168459703
035 $a AAI32455547
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Maharjan, Julina, $e author. $3 1010526
245 1 0 $a Transforming Social Media Data Into Mental Health Insights: Decoding Drug Abuse Patterns With Deep Learning and LLMs / $c Julina Maharjan.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2026
300 $a 1 electronic resource (202 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 87-06, Section: A.
500 $a Advisors: Jin, Ruoming Committee members: Kenne, Deric; Guan, Qiang; Maletic, Jonathan; Gandolfi, Enrico.
502 $b Ph.D. $c Kent State University $d 2026.
520 $a Online social media discourse data offers unprecedented opportunities to study drug use behaviors and related mental health signals at scale. This dissertation, Transforming Social Media Data into Mental Health Insights: Decoding Drug Abuse Patterns with LLMs, advances methodological frameworks that leverage deep learning, natural language processing (NLP), and large language models (LLMs) to decode complex, drug-related discourse on platforms such as Twitter/X.The first study implements large-scale deep learning pipelines to analyze drug/substance use patterns before, during, and after the COVID-19 pandemic. By processing millions of posts with a state-of-the-art Transformer-based language model (RoBERTa) and iterative Human-in-The-Loop fine-tuning, this study uncovers significant temporal shifts, particularly heightened alcohol discourse during pandemic peaks, and demonstrates how scalable NLP can map population-level health behaviors. The second study develops demographic-aware NLP models, integrating sentiment and emotion classifiers with demographic inference algorithms to capture nuanced differences across age, gender, and race in substance use discussions. These results were validated against national survey baselines, demonstrating how deep learning can generate insights of direct, actionable relevance to public health policy and interventions.Building on these core studies, the dissertation presents two additional methodological contributions. First, it provides a psychometric evaluation of LLM-derived embeddings for personality trait prediction, rigorously validating their reliability and interpretability against established psychological instruments. This work highlights that LLM embeddings sufficiently capture key emotion features, which are otherwise required in traditional personality prediction models. Second, it extends the framework to investigate the intersection of personality and substance use discourse during COVID-19, demonstrating the utility of personality-aware NLP models for behavioral health research.By uniting large-scale social media analysis with cutting-edge NLP and LLM methodologies, this dissertation establishes a robust computational foundation for decoding complex human language and behavior. The contributions demonstrate how methodological innovation; from scalable deep learning pipelines to theoretical advances in model adaptation, can transform raw digital traces into actionable insights for public health policy, targeted interventions, and the next generation of social media? based mental health research.
546 $a English
590 $a School code: 0101
650 4 $a Computer science. $3 199325
650 4 $a Web studies. $3 708690
653 $a Drug use
653 $a Personality psychology
653 $a Deep learning
653 $a Natural language processing
653 $a Mental health
690 $a 0984
690 $a 0800
690 $a 0646
710 2 $a Kent State University. $b College of Arts and Sciences / Department of Computer Science. $e degree granting institution. $3 1010527
720 1 $a Jin, Ruoming $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 87-06A.
790 $a 0101
791 $a Ph.D.
792 $a 2026
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32455547