Depression Detection Using Digital Traces on Social Media: A Knowledge-aware Deep Learning Approach

ABSTRACT

Depression is a pressing yet underdiagnosed issue in health management. Because depressed patients share their symptoms, life events, and treatments on digital platforms, information systems (IS) scholars resort to user-generated digital traces for depression detection. While they facilitate innovative information technology (IT) approaches to alleviate the social and economic burden of depression, most studies lack effective means to incorporate domain knowledge in depression detection systems or suffer from feature extraction difficulties. Following the design science research in IS, we propose a Deep-Knowledge-Aware Depression Detection system to detect social media users at risk of depression and explain the detection factors. We deploy extensive empirical analyses to evaluate our designed IT artifact, which shows domain knowledge greatly improves performance. Our work has significant implications for IS research in knowledge-aware machine learning, digital traces utilization, and generalizable design principles. Practically, the early detection and factor explanation from our IT artifact can assist depression management and enable large-scale assessment of the population’s mental health.

KEYWORDS:

• Design research
• healthcare analytics
• social media
• knowledge-aware deep learning
• depression
• online depression detection
• depression management

Supplementary Information

Supplemental data for this article can be accessed online at https://doi.org/10.1080/07421222.2024.2340822

Disclosure statement

No potential conflicts of interest are reported by the authors(s).

Notes

¹ A concept developed by the Harvard School of Public Health, WHO, and the World Bank, which is used to calculate disability, premature death, and other factors.

² When employees are present for work but less productive due to their illness.

³ A proper mechanism needs to consider expert-verified factors associated with depression, especially those frequently observable on social media. This mechanism also needs to contribute to an increase of F1-score in depression detection.

⁴ https://spacy.io/api/dependencyparser

⁵ The ontology used in the model is the same for all users. The entities in the ontology are the factors related to depression. Different users have different presentations of these factors. The ontology attention mechanism captures each user’s diverse similarity to each factor and uses such information in the prediction model.

⁶ Twitter reaches by age group 2021 in US: https://www.statista.com/statistics/265647/share-of-us-internet-users-who-use-twitter-by-age-group/.

Additional information

Notes on contributors

Wenli Zhang

Wenli Zhang ([email protected]; corresponding author) is an assistant professor in the Department of Information Systems and Business Analytics at Iowa State University’s Debbie and Jerry Ivy College of Business. Dr. Zhang’s interests revolve around the areas of data science and information system design, especially in developing techniques based on machine learning, deep learning, and natural language processing for solving real-world problems within the context of healthcare and other business concerns. Her prior works have been published in premier journals such as MIS Quarterly.

Jiaheng Xie

Jiaheng Xie ([email protected]) is an assistant professor in the Department of Accounting and MIS at Alfred Lerner College of Business and Economics, University of Delaware. His research interests lie in interpretable deep learning, health risk analytics, large language models, and business analytics. Dr. Xie’s prior works have been published in many premier journals, including Journal of Management Information Systems and MIS Quarterly.

Zhu (Drew) Zhang

Zhu Zhang ([email protected]) is the Verrecchia Endowed Chair in Artificial Intelligence and Business Analytics at the University of Rhode Island. His aspiration is to computationally discover valuable knowledge and intelligence from massive amounts of data available in various structures, genres, and streams. He builds AI-powered solutions to practical problems in a wide range of domains.

Xiang Liu

Xiang Liu ([email protected]) is a Ph.D. student at the Institute of Financial Services and Analytics, University of Delaware, researches deep learning, health risk, and business analytics. He explores innovative applications of these fields to revolutionize the healthcare industry.