Digital phenotyping of complex psychological responses to the COVID-19 pandemic

Abstract

Background: The novel coronavirus disease 2019 (COVID-19) has negatively impacted mortality, economic conditions, and mental health. A large scale study on psychological reactions to the pandemic to inform ongoing population-level symptom tracking and response to treatment is currently lacking.Methods: Average intake scores for standard depression and anxiety symptom scales were tracked from January 1, 2017 to June 9, 2020 for patients seeking treatment from a digital mental health service to gauge the relationship between COVID-19 and self-reported symptoms. We applied natural language processing (NLP) to unstructured therapy transcript data from patients seeking treatment during the height of the pandemic in the United States between March 1, 2020 and June 9, 2020 to identify words associated with COVID-19 mentions. This analysis was used to identify symptoms that were present beyond those assessed by standard depression and anxiety measures.Results: Depression and anxiety symptoms reported by 169,889 patients between January 1, 2017 and June 9, 2020 were identified. There was no detectable change in intake depression symptom scores. Intake anxiety symptom scores increased 1.42 scale points [95% CI: 1.18, 1.65] between March 15, 2020 and April 1, 2020, when scores peaked. In the transcript data of these 169,889 patients, plus an expanded sample of 49,267 patients without symptom reports, term frequency-inverse document frequency (tf-idf) identified 2,377 positively correlated and 661negatively correlated terms that were significantly (FDR<.01) associated with mentions of the virus. These terms were classifiable into 24 symptoms beyond those included in the diagnostic criteria for anxiety or depression.Conclusions: The COVID-19 pandemic may have increased intake anxiety symptoms for individuals seeking digital mental health treatment. NLP analyses suggest that standard symptom scales for depression and anxiety alone are inadequate to fully assess and track psychological reactions to the pandemic. Symptoms of grief, trauma, obsession-compulsion, agoraphobia, hypochondriasis, panic, and non- suicidal self-injury should be monitored as part of a new COVID-19 Syndrome category.