Multitask Learning for Mental Health Conditions with Limited Social Media Data

By Dirk Hovy, Margaret Mitchell, Adrian Donald Benton. 2017

149 citations recorded by the source corpus.

Language contains information about the author’s demographic attributes as well as their mental state, and has been successfully leveraged in NLP to predict either one alone. However, demographic attributes and mental states also interact with each other, and we are the first to demonstrate how to use them jointly to improve the prediction of mental health conditions across the board. We model the different conditions as tasks in a multitask learning (MTL) framework, and establish for the first time the potential of deep learning in the prediction of mental health from online user-generated text. The framework we propose significantly improves over all baselines and single-task models for predicting mental health conditions, with particularly significant gains for conditions with limited data. In addition, our best MTL model can predict the presence of conditions (neuroatypicality) more generally, further reducing the error of the strong feed-forward baseline.

Subfields: Artificial Intelligence, Social Psychology, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Social Psychology

In Natural Language Processing

Citations in the corpus

Anticipating Safety Issues in E2E Conversational AI: Framework and Tooling (2021, background): “Intentions of self harm Amongst the large body of literature on depression detection and mental health assessment in social media (e.g., Benton et al., 2017; Coppersmith et al., 2014; De Choudhury et al., 2013, inter alia), some research focuses on detecting risk of self-harm.”
Predictive Biases in Natural Language Processing Models: A Conceptual Framework and Overview (2019, background): “Recently, Benton et al. (2017) showed that accounting for demographic factors in the model could leverage this correlation and control for the confounds.”
Identifying beneficial task relations for multi-task learning in deep neural networks (2017, background): “Luong et al. (2016) suggest that it is important that the auxiliary data does not outsize the target data, while Benton et al. (2017) suggest that multi-task learning is particularly effective when we only have access to small amounts of target data.”
Disembodied Machine Learning: On the Illusion of Objectivity in NLP (2021, background): “models, multi-task learning models can encode information about the creator of the datum (Benton et al., 2017; Garcia et al., 2019) thus allowing models to take into account the embodiment of the datum.”
SafetyKit: First Aid for Measuring Safety in Open-domain Conversational Systems (2022, background): “Intentions of self harm Amongst the large body of work on mental health assessment in social media (e.g., Benton et al., 2017; Coppersmith et al., 2014; De Choudhury et al., 2013, inter alia), some research focuses on detecting risk of self-harm.”
Five sources of bias in natural language processing (2021, methodology): “NLP is used in industrial decision-making processes for hiring, abusive language and threat detection on social media (Roberts et al., 2019), and mental health assessment and treatment (Benton et al., 2017; Coppersmith et al., 2015).”
Using clinical Natural Language Processing for health outcomes research: Overview and actionable suggestions for future advances (2018, unknown): “Even in cases where an NLP method is used to classify a whole document (e.g., assigning tumor classifications to whole histopathology reports [61]), there may be several documents for an individual patient.”
Depression and Self-Harm Risk Assessment in Online Forums (2017, background): “…in social media to better discover and understand mental health related concerns (Resnik et al., 2013; De Choudhury et al., 2013; Copper-smith et al., 2014b,a; Mitchell et al., 2015; Tsug-awa et al., 2015; Coppersmith et al., 2015a; Al-thoff et al., 2016; Mowery et al., 2016; Benton et al., 2017b).”
Suicide Ideation Detection via Social and Temporal User Representations using Hyperbolic Learning (2021, methodology): “Although informed consent of each user was not sought as it may be deemed coercive, we perform automatic de-identification of the dataset using named entity recognition (Benton et al., 2017a,b) to reduce the risk of including any identifying data in the raw data.”
The Importance of Modeling Social Factors of Language: Theory and Practice (2021, background): “, 2017), or via neural models for multi-task learning (Benton et al., 2017; Li et al., 2018).”
Returning the N to NLP: Towards Contextually Personalized Classification Models (2020, methodology): “Benton et al. (2017) use age and gender as auxiliary tasks in a multitask learning setup for psychological labeling of users.”