Teaching Machines to Know Your Depressive State: On Physical Activity in Health and Major Depressive Disorder

Abstract

A less-invasive method for the diagnosis of the major depressive disorder can be useful for both the psychiatrists and the patients. We propose a machine learning framework for automatically discriminating patients suffering from the major depressive disorder (n n n = = = 1 1 14 4 4) and healthy subjects (n n n = = = 1 1 17 7 7). To this end, spontaneous physical activity data were recorded via a watch-type computer device equipped by the participants in their daily lives. Two machine learning models are investigated and compared, i. e., support vector machines, and deep recurrent neural networks. Experimental results show that, both of the two methods, i. e., the static model fed with human hand-crafted features, and the sequential model fed with raw data can reach a promising performance with an unweighted average recall at 76.0 % and 56.3 %, respectively.

Enhancing EEG-Based Classification of Depression Patients Using Spatial Information

“…The face-in-the-crowd task stimuli consisted of six human faces, which were selected from the Ekman emotion database [28]. There were three types of expressions (positive, negative, and neutral) without hair, glasses, beard, or other facial accessories.…”

“…Subject-independent k-fold cross-validation (CV) [24][25] and leave-one-subject-out (LOSO) CV [26][27][28] are two widely used EEG classification strategies. In fact, when k = 1, the LOSO method is a special case of the k-fold technique.…”

Can Appliances Understand the Behavior of Elderly Via Machine Learning? A Feasibility Study

“…In addition, for the static classifiers, e.g., a support vector machine (SVM) [47], we should extract superasegmental features [48] that are independent of the length of the analysed data. Motivated by our previous work [49], we applied nine statistical functionals successfully used in human behavior analysis. These functionals (see Fig.…”

“…Moreover, the current annotations are made by participants' self-reports, which maybe very subjective. In the future, we can consider using more reasonable and objective annotation methods as in our previous work on spontaneous physical analysis [49], [64], [82]. Accurately annotating the status of the elderly is a difficult, but essential work for ML-based methods.…”

Learning Multimodal Representations for Drowsiness Detection

“…The assumption of this study is that, the human behaviour in daily life can be modelled via the statistical functionals extracted from the spontaneous physical activity (SPA) and be used for predicting the drivers' drowsiness status. In our previous studies [33], [34], we found that using wearable sensor data collected from daily life can be efficiently exploited to analyse the human behaviour via machine learning methods. For instance, the SPA data recorded via a watch-type device can be used to detect if the subject is suffering from a major depressive disorder (MDD) [33].…”