Digital phenotyping for classification of anxiety severity during COVID-19

Abstract

COVID-19 has led to an increase in anxiety among Canadians. Canadian Perspectives Survey Series (CPSS) is a dataset created by Statistics Canada to monitor the effects of COVID-19 among Canadians. Survey data were collected to evaluate health and health-related behaviours. This work evaluates CPSS2 and CPSS4, which were collected in May and July of 2020, respectively. The survey data consist of up to 102 questions. This work proposes the use of the survey data characteristics to identify the level of anxiety within the Canadian population during the first- and second-phases of COVID-19 and is validated by using the General Anxiety Disorder (GAD)-7 questionnaire. Minimum redundancy maximum relevance (mRMR) is applied to select the top features to represent user anxiety, and support vector machine (SVM) is used to classify the separation of anxiety severity. We employ SVM for binary classification with 10-fold cross validation to separate the labels of Minimal and Severe anxiety to achieve an overall accuracy of 94.77±0.13% and 97.35±0.11% for CPSS2 and CPSS4, respectively. After analysis, we compared the results of the first and second phases of COVID-19 and determined a subset of the features that could be represented as pseudo passive (PP) data. The accurate classification provides a proxy on the potential onsets of anxiety to provide tailored interventions. Future works can augment the proposed PP data for carrying out a more detailed digital phenotyping.

Telehealth’s Role Enabling Sustainable Innovation and Circular Economies in Health

“…5,6 Digital phenotyping is an emerging DHI paradigm that relies on smartphones and wearables to support the continuity of care and improve scalability. [7][8][9][10][11][12][13][14][15] Multiple reviews have examined the positive evidence for effectiveness, cost-effectiveness, patient perceptions, and effects of telehealth on mortality. However, several ongoing concerns remain.…”

The Importance of Using Binary Classification Models in Predicting Depression from a Machine Learning Perspective

“…When analyzing minimal and mild versus moderate and severe using a GAD-7 score of 10 as a cut-off, the accuracy dropped to 87%-92%. The 4-class classification model achieved only 64%-74% accuracy[6]. Yue et al used internet traffic characteristics to classify whether one has depression or not using machine learning models and achieved an accuracy of 80%[7].…”