Exploring the Neuropsychiatric Sequalae of Perceived COVID-19 Exposure in College Students: A Pilot Digital Phenotyping Study

Abstract

The urgency to understand the long-term neuropsychiatric sequala of COVID-19, a part of the Post-Acute COVID-19 Syndrome (PACS), is expanding as millions of infected individuals experience new unexplained symptoms related to mood, anxiety, insomnia, headache, pain, and more. Much research on PACS involves cross sectional surveys which limits ability to understand the dynamic trajectory of this emerging phenomenon. In this secondary analysis, we analyzed data from a 4-week observational digital phenotyping study using the mindLAMP app for 695 college students with elevated stress who specified if they were exposed to COVID-19. Students also completed a biweekly survey of clinical assessments to obtain active data. Additionally, passive data streams like GPS, accelerometer, and screen state were extracted from phone sensors and through features the group built. Three hundred and eighty-second number participants successfully specified their COVID-19 exposure and completed the biweekly survey. From active smartphone data, we found significantly higher scores for the Prodromal Questionnaire (PQ) and the Pittsburgh Sleep Quality Index (PSQI) for students reporting exposure to COVID-19 compared to those who were not (ps < 0.05). Additionally, we found significantly decreased sleep duration as captured from the smartphone via passive data for the COVID-19 exposed group (p < 0.05). No significant differences were detected for other surveys or passive sensors. Smartphones can capture both self-reported symptoms and behavioral changes related to PACS. Our results around changes in sleep highlight how digital phenotyping methods can be used in a scalable and accessible manner toward better capturing the evolving phenomena of PACS. The present study further provides a foundation for future research to implement improving digital phenotyping methods.

Digital Phenotyping Data to Predict Symptom Improvement and App Personalization: Protocol for a Prospective Study

“…While many smartphone digital phenotyping biomarkers and algorithms have been proposed across the mental health field and even specifically for college mental health [ 11 ], none have ever been prospectively validated. In our prior research, we have been able to retrospectively validate findings, used to inform this study’s methods, on older data sets [ 6 , 8 ].…”

“…In addition, the app will be used to administer surveys and provide cognitive games, mindfulness, and other activities. Like earlier iterations of this study, college students will be recruited via social media to complete a screening survey on REDCap [ 8 ]. Given that in-person recruitment remains challenging around COVID-19, online recruitment via social media is practical [ 9 ].…”

“…To achieve our first aim, we present an additional logistic regression model to predict if participants will improve by at least 25% by the end of the study on the weekly surveys from the average of all features over the course of the study. The model was trained on data from the first iteration of the college study [ 8 ] and tested on the second iteration of the college study to test model generalization. The AUC scores are shown in Table 3 .…”