SHIWEN ZHOU

Ms. SHIWEN ZHOU

orcid

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
9 April 2025 in IEEE Access
Verified
Journal Article
Open access
SHIWEN ZHOUSHIWEN ZHOU
SHIWEN ZHOU
Masnizah MohdMasnizah Mohd
Masnizah Mohd
Social media expansion enables real-time mental health monitoring particularly for depression detection through its platforms. Traditional depression detection techniques, however, struggle to handle the informal, often ambiguous nature of language used in social media, including emojis, slang, and context-dependent expressions. To address this limitation, this research proposes a novel framework for depression detection that integrates advanced text preprocessing with a deep learning model. The preprocessing pipeline contains two key components where emojis are converted into text by emoji normalization then slang expressions are replaced through a customized dictionary. Emotion scores serve as additional components to identify depression-related emotional cues within the text. The model combines BERT’s contextual embeddings with BiLSTM’s sequential processing power to effectively represent emotional content in social media posts. The proposed approach demonstrates superior performance compared to traditional baseline models by achieving better results in accuracy, precision, recall, F1-score, AUC and RGA. This approach shows exceptional performance in detecting depression through informal texts which serves as a new benchmark for depression detection in dynamic social media environments.

18 May 2026 in Applied Sciences
Verified
Journal Article
Open access
PDF file
SHIWEN ZHOUSHIWEN ZHOU
SHIWEN ZHOU
Masnizah MohdMasnizah Mohd
Masnizah Mohd
LZLailatul Qadri Zakaria
Lailatul Qadri Zakaria
Depression is a critical global public health challenge, and the demand for accurate automated detection methods has generated considerable research interest in Transformer-based models. Despite their substantial promise, a comprehensive investigation into their architectural efficacy, intrinsic mechanisms, and barriers to practical implementation remains lacking. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines, this systematic review was conducted across six databases (IEEE Xplore, Elsevier, Springer, MDPI, PubMed, and arXiv). The final search was performed in October 2025, covering English-language empirical studies published between 2020 and 2025 that employed Transformer-based architectures for depression detection. Risk of bias and methodological quality were independently appraised by two authors using a six-dimension structured rubric, with disagreements resolved by a third author. Findings were narratively synthesized given substantial cross-study heterogeneity. This systematic review analyzed 46 studies and provided the first comprehensive, mechanism-level, architecturally stratified comparison of encoder-only, decoder-only, hybrid, and multimodal fusion paradigms, examining self-attention dynamics and transfer learning strategies. Since 2019, these frameworks have evolved from text-centric approaches to advanced multimodal systems. Encoder-only models show consistently strong results in high-throughput text-based screening, decoder-only models demonstrate stronger few-shot learning capabilities, hybrid architectures show the highest observed median performance in clinical interview settings across the reviewed studies, and multimodal fusion systems offer complementary advantages when heterogeneous signal integration is critical. These trends are task-contextualized and should not be interpreted as unconditional rankings, given heterogeneity in evaluation metrics and tasks across studies. Nonetheless, four principal challenges hinder clinical translation: overreliance on self-reported data, cross-linguistic bias, absence of uncertainty quantification, and substantial computational overhead. Future efforts should shift from incremental benchmark improvements toward clinical utility through standardized psychiatric validation, uncertainty-aware architectures, fairness-enforced training across diverse populations, and the integration of Transformer-based models with wearable and mobile health data to improve detection stability and reduce translational risk. This systematic review was registered on the Open Science Framework (OSF; DOI: 10.17605/OSF.IO/SYF9N). This research was funded by the Faculty of Information Science and Technology and by Universiti Kebangsaan Malaysia under Grant TAP-K014364.

13 November 2025
Verified
Conference Paper
Closed access
SHIWEN ZHOUSHIWEN ZHOU
SHIWEN ZHOU
MMMasnizah Mohd
Masnizah Mohd
LZLailatul Qadri Zakaria
Lailatul Qadri Zakaria
Social media analysis has become a critical field in computational mental health in detecting depression tendencies. This paper proposes an innovative Multi-Granularity SpatioTemporal Affective Perception Network (MSTAPN) that utilizes temporal dynamics and multi-scale feature extraction to achieve enhanced depression-detecting accuracy. Our approach integrates a specialized temporal attention mechanism with time-decay modeling, multi-granularity convolutions, and bidirectional LSTM networks to capture both short-term emotional fluctuations and long-term behavioral patterns. The method extracts 44 comprehensive features, comprising 24 static emotional indicators and 20 novel temporal characteristics such as posting patterns, emotional evolution trends, and time-interval distributions. Our MentalBERTMSTAPN model is evaluated on a balanced dataset of 498 users (249 depression, 249 control), achieving 0.6448 ROC-AUC and 0.6000 accuracy over the MentalBERT baseline. The temporal enhancement particularly excels in depression-specific detection with 0.6364 F1-score and 0.7000 recall. Cross-validation results confirm model robustness across data splits. The results indicate that temporal modeling of emotional expression patterns is a critical element toward effective depression detection using social media texts.