Authors:
(1) Yuchen Cao, Khoury college of computer science, Northeastern University;
(2) Jianglai Dai, Department of EECS, University of California, Berkeley;
(3) Zhongyan Wang, Center for Data Science, New York University;
(4) Yeyubei Zhang, School of Engineering and Applied Science, University of Pennsylvania;
(5) Xiaorui Shen, Khoury college of computer science, Northeastern University;
(6) Yunchong Liu, School of Engineering and Applied Science, University of Pennsylvania;
(7) Yexin Tian, Georgia Institute of Technology, College of Computing. Authors: Authors: (1) Yuchen Cao, Khoury college of computer science, Northeastern University; (2) Jianglai Dai, Department of EECS, University of California, Berkeley; (3) Zhongyan Wang, Center for Data Science, New York University; (4) Yeyubei Zhang, School of Engineering and Applied Science, University of Pennsylvania; (5) Xiaorui Shen, Khoury college of computer science, Northeastern University; (6) Yunchong Liu, School of Engineering and Applied Science, University of Pennsylvania; (7) Yexin Tian, Georgia Institute of Technology, College of Computing. Table of Links Abstract and Introduction
Methodology
Results
Discussion and Reference Abstract and Introduction Abstract and Introduction Methodology Methodology Results Results Discussion and Reference Discussion and Reference Abstract The global rise in depression necessitates innovative detection methods for early intervention. Social media provides a unique opportunity to identify depression through user-generated posts. This systematic review evaluates machine learning (ML) models for depression detection on social media, focusing on biases and methodological challenges throughout the ML lifecycle. A search of PubMed, IEEE Xplore, and Google Scholar identified 47 relevant studies published after 2010. The Prediction model Risk Of Bias ASsessment Tool (PROBAST) was utilized to assess methodological quality and risk of bias. Significant biases impacting model reliability and generalizability were found. There is a predominant reliance on Twitter (63.8%) and Englishlanguage content (over 90%), with most studies focusing on users from the United States and Europe. Non-probability sampling methods (approximately 80%) limit representativeness. Only 23% of studies explicitly addressed linguistic nuances like negations, crucial for accurate sentiment analysis. Inconsistent hyperparameter tuning was observed, with only 27.7% properly tuning models. About 17% did not adequately partition data into training, validation, and test sets, risking overfitting. While 74.5% used appropriate evaluation metrics for imbalanced data, others relied on accuracy without addressing class imbalance, potentially skewing results. Reporting transparency varied, often lacking critical methodological details. These findings highlight the need to diversify data sources, standardize preprocessing protocols, ensure consistent model development practices, address class imbalance, and enhance reporting transparency. By overcoming these challenges, future research can develop more robust and generalizable ML models for depression detection on social media, contributing to improved mental health outcomes globally. Introduction Depression and other mental health conditions have emerged as significant global health concerns, affecting over 264 million people worldwide, according to the World Health Organization (WHO, 2020). The increasing prevalence of mental health issues underscores the urgent need for effective detection and intervention strategies. Early identification of depression can lead to timely treatment and better outcomes, ultimately reducing the burden on patients, their caregivers, and healthcare systems (Kessler et al., 2017). In today’s digital age, social media platforms such as Twitter, Facebook, and Reddit play a central role in daily life for millions of people. These platforms not only facilitate communication but also serve as public outlets where individuals openly express their thoughts, emotions, and mental states (Choudhury et al., 2013). The extensive user-generated content on these platforms provides a unique opportunity for mental health research, enabling the real-time analysis of linguistic patterns and behavioral trends, and providing insights that may otherwise be inaccessible (Guntuku et al., 2017). Advancements in machine learning and deep learning have significantly enhanced the ability to process and analyze large-scale datasets. These technologies are particularly well-suited for handling the complex and nuanced data found on social media, as they can identify patterns and make predictions based on textual and behavioral cues. This capability makes them valuable tools for mental health detection, allowing researchers to develop models that can potentially identify at-risk individuals based on their social media activity (Shatte et al., 2019). By leveraging algorithms capable of learning from textual and behavioral cues, researchers are able to develop models that contribute to early intervention efforts in mental health care. Overview of Historical Studies on Machine Learning Approaches for Mental Health Detection in Social Media A growing body of research has explored the application of machine learning techniques to detect depression through social media analysis. Various algorithms—from traditional machine learning techniques such as logistic regression and support vector machines to advanced deep learning models and ensemble methods—have been employed to classify user posts and predict mental health conditions based on linguistic and semantic features (De Choudhury et al., 2013; Yazdavar et al., 2020). Platforms like Twitter, Facebook, and Reddit are frequently utilized due to their large user bases and the accessibility of public availability of data. One of the most common approaches within this research involves sentiment analysis, which aims to determine the emotional tone of user-generated content. By assessing positive, negative, or neutral sentiments expressed in posts, researchers attempt to correlate language patterns with indicators of depression (Kumar et al., 2020). For instance, increased usage of first-person singular pronouns and negative emotion words has been associated with depressive symptoms (Rude et al., 2004). Despite these promising results, multiple challenges persist. First, many studies suffer from limited generalizability due to small or homogeneous samples that may not represent the broader population. Data bias is a significant concern, stemming from the overrepresentation of certain demographic groups or linguistic communities while underrepresenting others (Olteanu et al., 2019). Moreover, a lack of robust sampling methods and standardized protocols impedes the reliability of findings. Third, the underutilization of advanced machine learning models, along with insufficient handling of complex linguistic nuances, such as sarcasm or context-dependent meanings, further restrict the effectiveness of these detection efforts (Calvo et al., 2017). Research Gaps and Objectives of the Current Study While individual studies have provided valuable insights into the machine learning application for mental health detection, there remains a noticeable lack of comprehensive reviews that consolidate the effectiveness of machine learning models across various studies. As discussed, current literature often falls short in addressing key methodological challenges throughout the entire lifecycle of machine learning and deep learning applications, including sampling, data preprocessing, model construction, and evaluation (Johnson et al., 2019). Although biases and limitations have been evaluated within some individual studies by their authors, their broader implications across all applications of machine learning and deep learning techniques in depression detection have not been fully explored. Therefore, a systematic review is essential to unify these findings and assess the pervasiveness and impact of biases across studies. To address these gaps, this study aims to conduct a systematic review that synthesizes and evaluates existing machine-learning models for detecting depression on social media. The specific objectives are: 1. Examine the effectiveness of machine learning and deep learning models by focusing on bias present in sampling, data preprocessing, model construction, fine-tuning, evaluation, and comparison, as well as the challenges associated with model generalizability across different social media platforms. 2. Explore methodological challenges, including those unique to mental health detection— such as handling class imbalances where depressive posts are the minority and preprocessing for sentiment analysis involving negations. Additionally, more general machine learning challenges, like improving model evaluation techniques and addressing data biases related to language and platform-specific factors, also persist. It is important to recognize that most of these biases arise unintentionally, either from practical challenges or from a lack of standardized guidelines for applying machine learning to mental health detection. By addressing these obstacles, the review aims to provide insights and strategies to mitigate these unintended biases, advancing the development of more reliable and generalizable models. 3. Provide recommendations for future research to enhance the reliability and applicability of machine learning models in mental health detection. These insights aim to inform strategies that improve early intervention efforts and contribute to the development of more robust, generalizable, and ethically sound machine learning applications. In doing so, the review seeks to provide guidance that fills the gap left by current practice, where a lack of formal guidelines has sometimes led to the persistence of unintentional biases. By addressing these objectives, this review seeks to provide a comprehensive understanding of the current practices and limitations within the field. The findings aim to guide future efforts toward developing more robust, generalizable, and ethically sound machinelearning applications for mental health detection using social media data. In the following sections, we will first examine the methodologies and models used across studies, followed by an analysis of common biases and limitations. We will conclude with a discussion on best practices and recommendations for advancing the field. This paper is available on arxiv under CC BY 4.0 license. This paper is available on arxiv under CC BY 4.0 license. available on arxiv

Part of HackerNoon's growing list of open-source research papers, promoting free access to academic material.

Facebook

AI, Depression & Social Media: What New Research Tells Us

Understanding the Role of Neuron Weights in AI Model Performance

MediaBias.TECH

Too Long; Didn't Read

Make resilience your competitive advantage

Can AI Tell When You’re Depressed?

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

Untitled Story

Bias in Machine Learning Models for Depression Detection

5 Ways AI is Radically Transforming the Healthcare Industry

6 Ways Emotion AI is Changing the Healthcare Industry

7 ways AI Revolutionizes Cancer Prediction, Detection, and Treatment

9 diagnostic aid AI products in 2021

Bias in Machine Learning Models for Depression Detection

5 Ways AI is Radically Transforming the Healthcare Industry

6 Ways Emotion AI is Changing the Healthcare Industry

7 ways AI Revolutionizes Cancer Prediction, Detection, and Treatment

9 diagnostic aid AI products in 2021

Light-Mode

Classic

Newspaper

Dark-Mode

Neon Noir

Minty

HN StartUps