Personalized Student Stress Prediction with Deep Multitask Network

Shaw, Abhinav; Simsiri, Natcha; Deznaby, Iman; Fiterau, Madalina; Tauhidur, Rahaman,

doi:10.48550/arxiv.1906.11356

Cited by 2 publications

(3 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They stated that if the number of epochs were increased, the training accuracy reached a maximum of 99%, but a drop in the performance of the testing set was noticed. The authors in [7][8][9][10][11][12] proposed a method to predict personalized stress in students using a deep multitask network. They received the dataset from a SmartLife study of 48 students in Dartmouth.…”

Section: Deep Learning Algorithmsmentioning

confidence: 99%

See 1 more Smart Citation

Enhancing Diagnostic Decision-Making: Ensemble Learning Techniques for Reliable Stress Level Classification

Anand,

Md,

Urooj

et al. 2023

Diagnostics

View full text Add to dashboard Cite

An intense level of academic pressure causes students to experience stress, and if this stress is not addressed, it can cause adverse mental and physical effects. Since the pandemic situation, students have received more assignments and other tasks due to the shift of classes to an online mode. Students may not realize that they are stressed, but it may be evident from other factors, including sleep deprivation and changes in eating habits. In this context, this paper presents a novel ensemble learning approach that proposes an architecture for stress level classification. It analyzes certain factors such as the sleep hours, productive time periods, screen time, weekly assignments and their submission statuses, and the studying methodology that contribute to stress among the students by collecting a survey from the student community. The survey data are preprocessed to categorize stress levels into three categories: highly stressed, manageable stress, and no stress. For the analysis of the minority class, oversampling methodology is used to remove the imbalance in the dataset, and decision tree, random forest classifier, AdaBoost, gradient boost, and ensemble learning algorithms with various combinations are implemented. To assess the model’s performance, different metrics were used, such as the confusion matrix, accuracy, precision, recall, and F1 score. The results showed that the efficient ensemble learning academic stress classifier gave an accuracy of 93.48% and an F1 score of 93.14%. Fivefold cross-validation was also performed, and an accuracy of 93.45% was achieved. The receiver operating characteristic curve (ROC) value gave an accuracy of 98% for the no-stress category, while providing a 91% true positive rate for manageable and high-stress classes. The proposed ensemble learning with fivefold cross-validation outperformed various state-of-the-art algorithms to predict the stress level accurately. By using these results, students can identify areas for improvement, thereby reducing their stress levels and altering their academic lifestyles, thereby making our stress prediction approach more effective.

show abstract

Section: Deep Learning Algorithmsmentioning

confidence: 99%

“…Performance of the stump = 1 2 log e 1 − Total Error Total Error (6) Then, the new weights after each iteration are updated using Equation (7).…”

Section: Stress Level Classification Using Adaboost Algorithmmentioning

confidence: 99%

Enhancing Diagnostic Decision-Making: Ensemble Learning Techniques for Reliable Stress Level Classification

Anand,

Md,

Urooj

et al. 2023

Diagnostics

View full text Add to dashboard Cite

show abstract

“…With the rise of modern machine learning and deep learning methods, these methods have been applied in the study of heart rate variability. Machine learning and deep learning methods have previously been used with HRV and electrocardiography (ECG) data for various applications such as: fatigue and stress detection [15]- [20], student stress prediction [21], congestive heart failure detection [17], [22], cardiac arrhythmia classification [23], [24]. The vast majority of prior arts, however, are supervised or…”

Section: Introductionmentioning

confidence: 99%

DeStress: Deep Learning for Unsupervised Identification of Mental Stress in Firefighters from Heart-Rate Variability (HRV) Data

Oskooei

Chau

Weiß

et al. 2020

Studies in Computational Intelligence

View full text Add to dashboard Cite

In this work we perform a study of various unsupervised methods to identify mental stress in firefighter trainees based on unlabeled heart rate variability data. We collect RR interval time series data from nearly 100 firefighter trainees that participated in a drill. We explore and compare three methods in order to perform unsupervised stress detection: 1) traditional K-Means clustering with engineered time and frequency domain features 2) convolutional autoencoders and 3) long short-term memory (LSTM) autoencoders, both trained on the raw RRI measurements combined with DBSCAN clustering and K-Nearest-Neighbors classification. We demonstrate that K-Means combined with engineered features is unable to capture meaningful structure within the data. On the other hand, convolutional and LSTM autoencoders tend to extract varying structure from the data pointing to different clusters with different sizes of clusters. We attempt at identifying the true stressed and normal clusters using the HRV markers of mental stress reported in the literature. We demonstrate that the clusters produced by the convolutional autoencoders consistently and successfully stratify stressed versus normal samples, as validated by several established physiological stress markers such as RMSSD, Max-HR, Mean-HR and LF-HF ratio.

show abstract

Personalized Student Stress Prediction with Deep Multitask Network

Cited by 2 publications

References 15 publications

Enhancing Diagnostic Decision-Making: Ensemble Learning Techniques for Reliable Stress Level Classification

Enhancing Diagnostic Decision-Making: Ensemble Learning Techniques for Reliable Stress Level Classification

DeStress: Deep Learning for Unsupervised Identification of Mental Stress in Firefighters from Heart-Rate Variability (HRV) Data

Contact Info

Product

Resources

About