Psychological precursors of epileptic seizures.

Psychological stress is the most commonly self-reported precursor of epileptic seizures. However, retrospective and prospective studies remain inconclusive in this regard. Here, we explored whether seizures would be preceded by significant changes in reported stressors or resource utilization. This study is based on high-frequency time series through daily online completion of personalized questionnaires of 9-24 items in epilepsy outpatients and compared responses 1-14 days before seizures with interictal time series. Fourteen patients (79% women, age = 23-64 years) completed daily questionnaires over a period of 87-898 days (median = 277 days = 9.2 months). A total of 4560 fully completed daily questionnaires were analyzed, 685 of which included reported seizure events. Statistically significant changes in preictal compared to interictal dynamics were found in 11 of 14 patients (79%) across 41 items (22% of all 187 items). In seven of 14 patients (50%), seizures were preceded by a significant mean increase of stressors and/or a significant mean decrease of resource utilization. This exploratory analysis of long-term prospective individual patient data on specific stressors and personal coping strategies generates the hypothesis that medium-term changes in psychological well-being may precede the occurrence of epileptic seizures in some patients.

Can Personalization Persuade? Study of Notification Adaptation in Mobile Behavior Change Intervention Application.

The growing ubiquity of smartphones and the ease of creating and distributing applications render the mobile platform an attractive means for facilitating positive behavior change at scale. Within the smartphone as a behavior change support system, mobile notifications play a critical role as they enable timely and relevant information distribution. In this paper we describe our preliminary investigation of the persuasiveness of mobile notifications delivered within a real-world behavior change intervention mobile app, which enabled users to set goals and define tasks related to those goals. The application aimed to motivate the users with notifications belonging to one of two groups-tailored and non-tailored, seeing them as sparks in the Fogg Behavior Model and personalizing them according to the users' Big Five personality traits. Results indicate that customized messages may work for some individuals while working poorly for others. When analyzing users as a single group, no significant differences were observed, but when proceeding with the analysis on the individual level we found seven users whose personality traits notifications interact with in interesting ways. Our results offer two general insights: (1) Using personality-tailored messaging in a dynamic mobile domain as opposed to a static domain leads to different outcomes, and it seems that there is no one-to-one mapping between domains; (2) A major reason for most of our hypotheses being false may be that messages that are deemed as persuasive on their own are not what persuades people to perform an action. Unlike the clear-cut findings observed in other domains, we discover a rather nuanced relationship between the personalization and persuasiveness that calls for further exploration at the individual participant level.

Machine Learning for Analyzing Non-Countermeasure Factors Affecting Early Spread of COVID-19.

The COVID-19 pandemic affected the whole world, but not all countries were impacted equally. This opens the question of what factors can explain the initial faster spread in some countries compared to others. Many such factors are overshadowed by the effect of the countermeasures, so we studied the early phases of the infection when countermeasures had not yet taken place. We collected the most diverse dataset of potentially relevant factors and infection metrics to date for this task. Using it, we show the importance of different factors and factor categories as determined by both statistical methods and machine learning (ML) feature selection (FS) approaches. Factors related to culture (e.g., individualism, openness), development, and travel proved the most important. A more thorough factor analysis was then made using a novel rule discovery algorithm. We also show how interconnected these factors are and caution against relying on ML analysis in isolation. Importantly, we explore potential pitfalls found in the methodology of similar work and demonstrate their impact on COVID-19 data analysis. Our best models using the decision tree classifier can predict the infection class with roughly 80% accuracy.