Abstract: Ego depletion refers to the idea that self-control and decision-making abilities become depleted or diminished after engaging in prolonged or demanding cognitive tasks. When conducting research with humans, states such as stress are usually asked directly to the participants by using ecological momentary assessment (EMA). However, as those answers are self-reported, they are prone to bias due to the lack of user commitment or effort to assess the stress levels the best they can. This paper investigates the relationship between ego depletion and stress level reporting, specifically focusing on the bias in self-reported stress levels. The hypothesis tested in this study suggests that higher levels of ego depletion lead to greater bias in self-reported stress levels. Data collected by Berrocal et al. using EMA were analyzed to examine this hypothesis. The dataset included self-reporting stress levels by individuals being assessed and stress level reports from their peers. Data was collected using a mobile app, incorporating passive smartphone usage data alongside EMA responses. The hypothesis was tested and studied employing artificial intelligence algorithms. The results partially confirmed the initial hypothesis, revealing that lower ego depletion was associated with reduced bias in stress level reporting. Notably, when analyzing data from working days, the morning period demonstrated the least bias compared to the afternoon. These findings suggest that individuals have higher self-capacity and willingness to provide accurate stress level assessments during the morning hours. Challenges encountered in the research included limitations related to the holidays considered and potential confounders such as flexible time schedules or post-lunch sleepiness. This paper releases its data publicly, allowing for further examination and replication of the findings. Future research is encouraged to expand upon these conclusions.

Abstract: Wearable devices are a useful and widely used source of continuous and temporal dependant data. In contrast to the traditional clinical environment, these devices allow time series data collection in an individual’s daily living environ- ment. However, missing data can occur while using them. Many techniques have been applied to solve these data gaps; nonetheless, missing time series data poses extra challenges, such as maintaining the temporal dependency. In this article, we addressed the forecast of sleep trackers data (sleeping heart rate (HR) and time asleep) for 2 main reasons: (1) to design models capable of accurately forecasting missing data from those devices, and (2) to apply those models to empower sleep interventions that may increase its quality, by forecasting future sleep events. We collected wearables data over 290 days (per individual) from 12 participants using a smartwatch and made this dataset publicly available. We then explored several hyperparameters of 2 Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU). We further elaborated and compared the performance of 3 approaches to training those RNNs. Although similar performance, slightly more accurate results were obtained after training a GRU network on an entire population’s dataset, which was able to forecast the average, minimum, and maximum sleeping HR with a root-mean-squared error (RMSE) of 4.4 (± 1.4), 4.9 (± 2.6), and 12.1 (± 4.0) beats per minute, respectively. However, the total time asleep was impossible to forecast with low error.

Abstract: Wearable devices are a useful and widely used source of continuous and temporal dependant data. In contrast to the traditional clinical environment, these devices allow time series data collection in an individual’s daily living environ- ment. However, missing data can occur while using them. Many techniques have been applied to solve these data gaps; nonetheless, missing time series data poses extra challenges, such as maintaining the temporal dependency. In this article, we addressed the forecast of sleep trackers data (sleeping heart rate (HR) and time asleep) for 2 main reasons: (1) to design models capable of accurately forecasting missing data from those devices, and (2) to apply those models to empower sleep interventions that may increase its quality, by forecasting future sleep events. We collected wearables data over 290 days (per individual) from 12 participants using a smartwatch and made this dataset publicly available. We then explored several hyperparameters of 2 Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU). We further elaborated and compared the performance of 3 approaches to training those RNNs. Although similar performance, slightly more accurate results were obtained after training a GRU network on an entire population’s dataset, which was able to forecast the average, minimum, and maximum sleeping HR with a root-mean-squared error (RMSE) of 4.4 (± 1.4), 4.9 (± 2.6), and 12.1 (± 4.0) beats per minute, respectively. However, the total time asleep was impossible to forecast with low error.