Development and validation of a machine learning-based vocal predictive model for major depressive disorder

From smartphone data to clinically relevant predictions: A systematic review of digital phenotyping methods in depression

“…Cao, et al (2020) found that a higher depression score is significantly correlated with lower social interaction levels (i.e., shorter phone call durations, fewer text messages sent), which seems consistent with Sverdlov, et al’s (2021) findings from app-based investigations, despite the differences in data type. Wasserzug, et al (2023) also found a significant correlation between vocal depression scores derived from prosodic vocal features and depressive symptom scores.…”

“…Studies were grouped by analysis goal to allow for comparison of methods with similar objectives. To this end, to investigate our first research question we first compared studies that correlated individual passive smartphone features with depression symptom severity (Cao, et al, 2020; Sun, et al, 2023; Sverdlov, et al, 2021; Wasserzug, et al, 2023; Zhang, et al, 2022; Zou, et al, 2023). To address our second research question, we then shifted our focus to the different methods that have been used for various depression prediction tasks, investigating the methods used for predicting symptom severity (Braund, et al, 2022; Cao, et al, 2020; Faurholt-Jepsen, et al, 2022; Kathan, et al, 2022; Pedrelli, et al, 2020; Pellegrini, et al, 2022; Sverdlov, et al, 2021; Zhang, et al, 2021; Zhang, et al, 2022).…”

“…Somewhat unexpectedly, two studies aimed to predict specific smartphone features from ratings of depression (Laiou, et al, 2022; Tønning, Faurholt-Jepsen, Frost, Bardram, & Kessing, 2021). We then shifted our focus towards studies that aimed to classify participants into different diagnostic classes and mood states (Bai, et al, 2021; Cho, et al, 2019; Faurholt-Jepsen, et al, 2022; Lee, et al, 2023; Kim, et al, 2023; Sverdlov, et al, 2021; Wasserzug, et al, 2023). Some studies contained unique goals that were not shared with the other studies, and these goals are considered in a separate section.…”

“…Wasserzug, et al (2023) classified MDD patients (in acute depressive states and in remission) and healthy controls into ‘high depression’ and ‘low depression’ states using a model developed from vocal features that were correlated with depression, achieving a cross-validation accuracy of 72.7%.…”

“…The success rate of classification was found to vary depending on which classes and features were included in each binary classifier. The set of features that generallyWasserzug, et al (2023) classified MDD patients (in acute depressive states and in remission) and healthy controls into 'high depression' and 'low depression' states using a model developed from vocal features that were correlated with depression, achieving a crossvalidation accuracy of 72.7%.Random forest models were used byCho, et al (2019) to classify mood state as biased or neutral, and mood episode as depressive, manic, hypomanic, or no episode. Behavioural patterns from participants with bipolar disorder were also investigated Cho, et al (2019).…”