Identifying Depressive Symptoms from Tweets: Figurative Language Enabled Multitask Learning Framework

Existing studies on using social media for deriving mental health status of users focus on the depression detection task. However, for case management and referral to psychiatrists, healthcare workers require practical and scalable depressive disorder screening and triage system. This study aims to design and evaluate a decision support system (DSS) to reliably determine the depressive triage level by capturing fine-grained depressive symptoms expressed in user tweets through the emulation of Patient Health Questionnaire-9 (PHQ-9) that is routinely used in clinical practice. The reliable detection of depressive symptoms from tweets is challenging because the 280-character limit on tweets incentivizes the use of creative artifacts in the utterances and figurative usage contributes to effective expression. We propose a novel BERT based robust multi-task learning framework to accurately identify the depressive symptoms using the auxiliary task of figurative usage detection. Specifically, our proposed novel task sharing mechanism, co-task aware attention, enables automatic selection of optimal information across the BERT layers and tasks by soft-sharing of parameters. Our results show that modeling figurative usage can demonstrably improve the model's robustness and reliability for distinguishing the depression symptoms.

[1]  Cecile Paris,et al.  Figurative Usage Detection of Symptom Words to Improve Personal Health Mention Detection , 2019, ACL.

[2]  Mark Dredze,et al.  Discovering Shifts to Suicidal Ideation from Mental Health Content in Social Media , 2016, CHI.

[3]  Martin D. Sykora,et al.  What about Mood Swings: Identifying Depression on Twitter with Temporal Measures of Emotions , 2018, WWW.

[4]  Sriparna Saha,et al.  Assessing the Severity of Health States based on Social Media Posts , 2020, 2020 25th International Conference on Pattern Recognition (ICPR).

[5]  James W. Pennebaker,et al.  Linguistic Inquiry and Word Count (LIWC2007) , 2007 .

[6]  L. Gottschalk Language in Psychiatry: A Handbook of Clinical Practice , 2007 .

[7]  Ming-Wei Chang,et al.  BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding , 2019, NAACL.

[8]  Mohammad Soleymani,et al.  AVEC 2019 Workshop and Challenge: State-of-Mind, Detecting Depression with AI, and Cross-Cultural Affect Recognition , 2019, AVEC@MM.

[9]  Vidhyasaharan Sethu,et al.  Analysis of acoustic space variability in speech affected by depression , 2015, Speech Commun..

[10]  J. Mundt,et al.  Vocal Acoustic Biomarkers of Depression Severity and Treatment Response , 2012, Biological Psychiatry.

[11]  Elliot Moore,et al.  Critical Analysis of the Impact of Glottal Features in the Classification of Clinical Depression in Speech , 2008, IEEE Transactions on Biomedical Engineering.

[12]  M. Haselton,et al.  The Evolution of Cognitive Bias , 2015 .

[13]  Dirk Hovy,et al.  Multitask Learning for Mental Health Conditions with Limited Social Media Data , 2017, EACL.

[14]  Albert A. Rizzo,et al.  Automatic audiovisual behavior descriptors for psychological disorder analysis , 2014, Image Vis. Comput..

[15]  Elmar Nöth,et al.  Automatic modelling of depressed speech: relevant features and relevance of gender , 2014, INTERSPEECH.

[16]  Mark Dredze,et al.  Shared Task : Depression and PTSD on Twitter , 2015 .

[17]  Nazli Goharian,et al.  Depression and Self-Harm Risk Assessment in Online Forums , 2017, EMNLP.

[18]  Eric Horvitz,et al.  Characterizing and predicting postpartum depression from shared facebook data , 2014, CSCW.

[19]  Jiasen Lu,et al.  Hierarchical Question-Image Co-Attention for Visual Question Answering , 2016, NIPS.

[20]  Fabien Ringeval,et al.  AVEC 2017: Real-life Depression, and Affect Recognition Workshop and Challenge , 2017, AVEC@ACM Multimedia.

[21]  Mohammad H. Mahoor,et al.  Nonverbal social withdrawal in depression: Evidence from manual and automatic analyses , 2014, Image Vis. Comput..

[22]  David A. Clifton,et al.  Detecting Adolescent Psychological Pressures from Micro-Blog , 2014, HIS.

[23]  David DeVault,et al.  The Distress Analysis Interview Corpus of human and computer interviews , 2014, LREC.

[24]  R. Spitzer,et al.  The PHQ-9: A new depression diagnostic and severity measure , 2002 .

[25]  Björn W. Schuller,et al.  AVEC 2011-The First International Audio/Visual Emotion Challenge , 2011, ACII.

[26]  Eric Horvitz,et al.  Predicting postpartum changes in emotion and behavior via social media , 2013, CHI.

[27]  Tingshao Zhu,et al.  Detecting Suicidal Ideation in Chinese Microblogs with Psychological Lexicons , 2014, 2014 IEEE 11th Intl Conf on Ubiquitous Intelligence and Computing and 2014 IEEE 11th Intl Conf on Autonomic and Trusted Computing and 2014 IEEE 14th Intl Conf on Scalable Computing and Communications and Its Associated Workshops.

[28]  Roland Göcke,et al.  An Investigation of Depressed Speech Detection: Features and Normalization , 2011, INTERSPEECH.

[29]  Amit P. Sheth,et al.  Multi-Task Learning Framework for Mining Crowd Intelligence towards Clinical Treatment , 2018, NAACL.

[30]  Qiang Yang,et al.  An Overview of Multi-task Learning , 2018 .

[31]  Rupa S. Valdez,et al.  Ethics in Health Research Using Social Media , 2019, Social Web and Health Research.

[32]  Björn W. Schuller,et al.  AVEC 2013: the continuous audio/visual emotion and depression recognition challenge , 2013, AVEC@ACM Multimedia.

[33]  Aron Halfin,et al.  Depression: the benefits of early and appropriate treatment. , 2007, The American journal of managed care.

[34]  Munmun De Choudhury,et al.  Modeling and Understanding Visual Attributes of Mental Health Disclosures in Social Media , 2017, CHI.

[35]  Pushpak Bhattacharyya,et al.  A Unified Multi-task Adversarial Learning Framework for Pharmacovigilance Mining , 2019, ACL.

[36]  Martial Hebert,et al.  Cross-Stitch Networks for Multi-task Learning , 2016, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[37]  Dipanjan Das,et al.  BERT Rediscovers the Classical NLP Pipeline , 2019, ACL.

[38]  Rafael A. Calvo,et al.  CLPsych 2016 Shared Task: Triaging content in online peer-support forums , 2016, CLPsych@HLT-NAACL.

[39]  George Kurian,et al.  Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation , 2016, ArXiv.

[40]  Pushpak Bhattacharyya,et al.  Medical Sentiment Analysis using Social Media: Towards building a Patient Assisted System , 2018, LREC.

[41]  Sebastian Ruder,et al.  An Overview of Multi-Task Learning in Deep Neural Networks , 2017, ArXiv.

[42]  Abdenour Hadid,et al.  Combining Global and Local Convolutional 3D Networks for Detecting Depression from Facial Expressions , 2019, 2019 14th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2019).

[43]  Ritu Garg,et al.  Multi-level Attention Network using Text, Audio and Video for Depression Prediction , 2019, AVEC@MM.

[44]  Manuel Montes-y-Gómez,et al.  Detecting Depression in Social Media using Fine-Grained Emotions , 2019, NAACL.

[45]  George Trigeorgis,et al.  End-to-End Multimodal Emotion Recognition Using Deep Neural Networks , 2017, IEEE Journal of Selected Topics in Signal Processing.

[46]  Dirk Hovy,et al.  The Social Impact of Natural Language Processing , 2016, ACL.

[47]  Timothy Miller,et al.  A BERT-based Universal Model for Both Within- and Cross-sentence Clinical Temporal Relation Extraction , 2019, Proceedings of the 2nd Clinical Natural Language Processing Workshop.

[48]  Fabien Ringeval,et al.  AVEC 2018 Workshop and Challenge: Bipolar Disorder and Cross-Cultural Affect Recognition , 2018, AVEC@MM.

[49]  Mark Dredze,et al.  Quantifying Mental Health Signals in Twitter , 2014, CLPsych@ACL.

[50]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[51]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[52]  Amit P. Sheth,et al.  Semi-Supervised Approach to Monitoring Clinical Depressive Symptoms in Social Media , 2017, ASONAM.

[53]  Leonardo Max Batista Claudino,et al.  Beyond LDA: Exploring Supervised Topic Modeling for Depression-Related Language in Twitter , 2015, CLPsych@HLT-NAACL.

[54]  Thomas F. Quatieri,et al.  A review of depression and suicide risk assessment using speech analysis , 2015, Speech Commun..

[55]  Munmun De Choudhury,et al.  Mental Health Discourse on reddit: Self-Disclosure, Social Support, and Anonymity , 2014, ICWSM.

[56]  Sriparna Saha,et al.  Relation Extraction From Biomedical and Clinical Text: Unified Multitask Learning Framework , 2020, IEEE/ACM Transactions on Computational Biology and Bioinformatics.

[57]  Albert A. Rizzo,et al.  Automatic behavior descriptors for psychological disorder analysis , 2013, 2013 10th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition (FG).

[58]  Roland Göcke,et al.  Diagnosis of depression by behavioural signals: a multimodal approach , 2013, AVEC@ACM Multimedia.