Search

Recognizing human cognitive performance is important for preserving working efficiency and preventing human error. This paper presents a method for estimating cognitive performance by leveraging multiple information available in a smartphone. The method employs the Go-NoGo task to measure cognitive performance, and fuses contextual and behavioral features to identify the level of performance. It was confirmed that the proposed method could recognize whether cognitive performance was high or low with an average accuracy of 71%, even when only referring to inertial sensor logs. Combining sensing modalities improved the accuracy up to 74%.

In the free energy principle (FEP) proposed by Friston, it is supposed that agents seek to minimize the “surprise” – the negative log (marginal) likelihood of observations (i.e., sensory stimuli) – given the agents’ current belief. This is achieved by minimizing the free energy, which provides an upper bound on the surprise. The FEP has been applied to action selection in a framework called “active inference,” where agents are supposed to select an action so that they minimize the “expected free energy” (EFE). While EFE can be decomposed into interpretable components such as epistemic value and extrinsic value, it is difficult to understand intuitively how EFE itself is directly related to “surprise” and what psychological construct is related to EFE itself (as a single quantity). To facilitate the discussion and interpretation of psychological processes underlying active inference, we introduce a computational component termed the “retrospective surprise,” which is the surprise of an observation after updating the belief given the observation itself. The predicted retrospective surprise (PRS) is mathematically derived from a special case of EFE and provides a lower bound on EFE. We illustrate the properties of EFE and PRS using examples of inference for a binary hidden cause given a binary observation. We discuss how information-seeking behavior is accounted for by EFE and PRS. Our results highlight the role of prior distribution of future observation in EFE. By setting this prior distribution to be fixed irrespective of which action is selected, the epistemic value can exert an influence on EFE, leading to information-seeking behavior., ファイル公開：2022-06-01

Recently, inferring the state of people’s mental health via passive mobile sensing has attracted significant attention. Previous studies have used the self-assessed stress levels as the ground truth; however, these are subjective measures. In this study, we use a physiologically-assessed stress metric to minimize the effect of participant subjectivity and further estimate it using behavioral features based on the smartphone usage logs. We initially requested the study participants (39 participants) to attach heart rate sensors for 8 hours per day and simultaneously collected continuous heart rate data and smartphone logs for 42 days. Further, we divided the participants into four types via clustering using the behavioral features derived from their smartphone sensor logs and trained each model via supervised learning using the heart rate data as the ground truth. Our results exhibit that the proposed method is more accurate (71%) as compared to the baseline method (54%). This demonstrates that physiologically-assessed stress levels can be estimated based on the implicit features that are gathered from the smartphone logs.

Mental health has received increasing attention in recent years, and a correlation between social interaction during eating and mental health noted. The purpose of this research is to determine if wearable devices can be used to recognize whether a person is eating alone or has company. Using a watch-type device and smartphone, the data, while eating in daily life, was collected. Using this data, we were able to recognize whether a person is eating alone or has company with an accuracy of 96.3%.

This study predicts the change of stress levels using real-world and online behavioral features extracted from smartphone log information. Previous studies of stress detection using smartphone data focused on a single feature and did not consider all features simultaneously. We propose a method to extract a co-occurring combination of a user's real-world and online behavioral features by converting raw sensor data into categorical features. We conducted an experiment in which the State Trait Anxiety Inventory (STAI) was used to assess the anxiety-related stress levels of 20 healthy participants. The participants installed a log-collecting application on their smartphones and answered the STAI questions once a day for one month. The proposed method showed an F-score of 74.2%, which is 4.0% higher than the F-score of previous studies (70.2%) that used single non-combined features. The results demonstrate that anxiety-related stress levels can be predicted using combined features extracted from smartphone log data.

The Sense of Agency (SoA) is the subjective sense such that I am the causal factor of the own experience [1].