A digital phenotyping dataset for impending panic symptoms: a prospective longitudinal study

Abstract

This study investigated the utilization of digital phenotypes and machine learning algorithms to predict impending panic symptoms in patients with mood and anxiety disorders. A cohort of 43 patients was monitored over a two-year period, with data collected from smartphone applications and wearable devices. This research aimed to differentiate between the day before panic (DBP) and stable days without symptoms. With RandomForest, GradientBoost, and XGBoost classifiers, the study analyzed 3,969 data points, including 254 DBP events. The XGBoost model demonstrated performance with a ROC-AUC score of 0.905, while a simplified model using only the top 10 variables maintained an ROC-AUC of 0.903. Key predictors of panic events included evaluated Childhood Trauma Questionnaire scores, increased step counts, and higher anxiety levels. These findings indicate the potential of machine learning algorithms leveraging digital phenotypes to predict panic symptoms, thereby supporting the development of proactive and personalized digital therapies and providing insights into real-life indicators that may exacerbate panic symptoms in this population.

Fig. 1

Data collection protocol.

Fig. 2

Overview of inPHRsym dataset.

Fig. 3

Study design and workflow of participants selection.

Fig. 4

Receiver operating characteristic curve (ROC) of panic prediction model using XGBoost classifier. The confidence band shows one standard deviation above and below the mean ROC curve of 5-fold cross-validation. Abbreviation: ROC, Receiver operating characteristic curve; AUC, Area under the receiver operating characteristic curve; CI, Confidence Interval.

Fig. 5

(A) The top 20 variables with the highest Shapley values from the XGBoost classifier trained on the entire digital phenotypic dataset and (B–G) box plots displaying the distribution of the 6 variables with the highest Shapley values for DBP and SD. Abbreviations: ‘(C)’, Clinical psychological data; ‘(D)’, Daily log data; ‘(S)’, Socio-demographic data; ‘(L)’, Life log data; CTQ, Childhood Trauma Questionnaire; STAI-X2, State-Trait Anxiety Inventory score – Trait Anxiety section; BRIAN, Biological Rhythms Interview of Assessment in Neuropsychiatry; KRQ-53, Korean Resilience Quotient-53; SPAQ, Seasonal Pattern Assessment Questionnaire; MDQ, Mood Disorder Questionnaire; CSM, Composite Scale of Morningness; SD, the stable day; DBP, the day before panic; ns, statistically no significant; ∗p < 0·05; ∗∗p < 0·01; ∗∗∗p < 0·001.

Fig. 6

Receiver operating characteristic curve (ROC) of entire digital phenotypic data model, daily log data model, life log data model, clinical data model, and Top-10-Shapley-data model. The confidence band shows one standard deviation above and below the mean ROC curve of 5-fold cross-validation. Abbreviation: ROC, Receiver operating characteristic curve; AUC, Area under the receiver operating characteristic curve; CI, Confidence Interval.

Fig. 7

Beeswarm plots of sample-specific Shapley values in the entire digital phenotypic data model (A), clinical data model (B), daily log data model (C), and life log data model (D). In these plots, each points represent the Shapley value of a particular variable for an individual sample. A wider distribution of points indicates the higher importance of corresponding variable in the model. Positive Shapley values, positioned on the right side of the plot, indicate variables that contribute to the prediction of a DBP, whereas negative Shapley values, located on the left side, correspond to variables that are predictive of SD. Abbreviations:: ‘(C)’, Clinical psychological data; ‘(D)’, Daily log data; ‘(S)’, Socio-demographic data; ‘(L)’, Life log data; CTQ, Childhood Trauma Questionnaire; STAI-X2, State-Trait Anxiety Inventory score – Trait Anxiety section; BRIAN, Biological Rhythms Interview of Assessment in Neuropsychiatry; KRQ-53, Korean Resilience Quotient-53; SPAQ, Seasonal Pattern Assessment Questionnaire; MDQ, Mood Disorder Questionnaire; CSM, Composite Scale of Morningness; SD, the stable day; DBP, the day before panic.