Natural Language Processing Of Helpline Chat Data Before And During The Pandemic Revealed Significant Decrease In Self-image Appreciation And Changes In Other Traits

Abstract

During the last two years the COVID-19 pandemic has affected the world population in several ways. An important increase in mental health problems is a consequence of this pandemic that is ubiquitous worldwide. In this work we study the effect of the pandemic on the mental health of a population of teenagers and youth based on the analysis of natural language processing, machine learning algorithms and expert knowledge. The data analysed was obtained from a chat helpline called Safe time from theIt Get’s Better Foundation in Chile. The data consists of 10,986 conversations gathered from 2018 until 2020 between volunteers from the foundation and users of the platform. We compared the conversationsbefore and during the pandemic in terms of their thematic content. Our analysis found: a significantdecrease in self-image appreciation during the pandemic; a significant decrease in the quality of personalrelationships during the pandemic, and a significant increase of performance appreciation.

Predicting Recurrent Chat Contact in a Low-Threshold Psychological Intervention for Children and Young Adults: A Natural Language Processing Approach

“…However, notably those reviews did not focus on text-only approaches and used a diverse set of predictors. The limited evidence from the same setting, chat-based counseling services, mostly reported very high predictive performances, such as an AUC of 0.9 for distress classification 27 and recall scores of close to 0.9 for detecting suicidal crisis 28 . While impressive in their domain, we argue that those are not good baselines of comparison for the prediction of outcomes like recurrent chat contact.…”

“…Here, the model reached high performances of AUC scores of close to 0.9, however the used test set did just include 78 chats. Another approach used around 5500 chat consultations to detect disclosure of suicidal ideation, using a deep-learning based model incorporating external domain knowledge 28 . Other work used NLP techniques for hypothesis-driven investigations like topical changes over the pandemic 29 or the development of chatbots 30,31 .…”

Modern views of machine learning for precision psychiatry

“…Aside from developing models for detecting suicidal ideation, 213 NLP can also be applied to these datasets to identify the population-level trend, such as the increase in anxiety and decrease in quality of personal relationships during the COVID-19 pandemic. 214 Since language data are ubiquitous, one of the NLP challenges in mental health applications is data standardization. Depending on the task, different types of data may yield different levels of “signal.” For example, to predict first-episode psychosis, language data from clinical tests have higher performance compared with transcripts of free speech.…”

Modern Views of Machine Learning for Precision Psychiatry

“…The advances in text-based mental health interventions (e.g., Talkspace and CrisisTextLine) have made transcripts of clinical sessions easily amenable via NLP. Aside from developing models for detecting suicide ideation [203], NLP can also be applied to these datasets to understand populationlevel trends, such as the increase in anxiety and decrease in quality of personal relationships during the COVID-19 pandemic [204]. Since language data are ubiquitous, one of the challenges in applying NLP for advancing mental health is standardizing data sources.…”