Psychological Stress Prediction of Human Using
Machine Learning Algorithms
Presenter
M. Raihan
Outlines
❑ Introduction
❑ Problem Statement
❑ Related Works
❑ Objectives
❑ Outcomes and Impacts
❑ Methodology
❑ Simulation Result
❑ Conclusion
❑ Future Research Direction
❑ Acknowledgement
❑ References
14 October 2025 2
Introduction
Depression has been one of the most common global
health issues for the last few decades.
Depression is a widespread disorder of the brain and
nervous system
This mental health condition currently affects
approximately 322 million individuals globally [4].
Machine learning (ML) techniques are increasingly used to
find significant patterns in various data sectors due to the
rapid growth of information and communication
technologies.
14 October 2025 3
Problem Statement
To anticipate psychological stress situations, the
primary goal of this research is to identify the
most significant stress features in various
individuals and apply several well-known
machine techniques.
14 October 2025 4
Related Works
Md. Sabab Zulfiker et al. [19] tested :
▪ Six ML classifiers using sociodemographic and
psychological data to predict depression.
▪ The SelectKBestand AdaBoost classifiers have 92.56%
accuracy.
▪ Multiple depression prediction models were
evaluated for F1-score, AUC, sensitivity, accuracy, and
specificity.
14 October 2025 5
Related Works (Continue)
Yu Xin et al. (2022) [18] used:
▪ The Random Forest classifier method to predict
depression in disabled seniors.
▪ Family, social, health, and demographics were evaluated.
▪ Rural, crippled, old, and sad adults had a 0.71
AUC,
65.3% specificity, and 80.6% sensitivity. Urban
crippled elderly sad people have a 0.78 AUC, 78.1%
specificity, and 64.2% sensitivity.
14 October 2025 6
Related Works (Continue)
Sunday O. Olatunji et al. [12] proposed:
▪ A clinically effective and economically efficient ML strategy
for early MS identification using clinical laboratory
biomarkers.
▪ SVM, DT, LR, and RF were compared.
▪ Following SelectKBest implementation, the ET classifier
used 11 features to achieve 94.74% Accuracy, 94.67%
precision, 95.95% F1-score, and 97.26% recall.
14 October 2025 7
Objectives
To extract the most significant features from the desired
dataset.
Several machine learning models were used, and we
picked the best one to predict stress conditions with the
help of the Burns Depression Checklist (BDC).
To conduct a comparative analysis and verify the findings
through a comprehensive review of relevant scholarly
literature.
14 October 2025 8
Outcomes and Impacts
❑We used fifteen features as like ENVSAT, INSOM, INFER,
POSSAT, ANXI, SUICIDE, ABUSED, CONFLICT, FINSTR, DEPRI,
THREAT, LOST, DEBT, EATDIS, CHEAT.
❑Precision, recall, F-Measure, ROC area, Kappa Statistic, and
accuracy are measured.
❑Four different classifiers are used: KNN, Addaboost,
Bagging, and SGD.
❑The Adaboost classifier’s reported accuracy is 90.81%,
which yields lower accuracy than other approaches. The
SGD classifier has the highest accuracy of 93.83%.
14 October 2025 9
Methodology
14 October 2025 10
Methodology (continue)
Dataset Collection
Data Visualization
Analysis Category
Data Training
Classification Algorithms
14 October 2025 11
Dataset Collection
▪ A survey was conducted to gather data from Bangladesh's
citizens of various age groups [19].
▪ A survey instrument comprising 55 items was developed. The
initial 30 questions collected psychosocial and socio-
demographic data from individuals, while the last 25 were from
the Burns Depression Checklist (BDC).
▪ A total of 604 individuals contributed to the dataset.
▪ Each symptom’s severity is added to calculate an individual’s
BDC score.
▪ The BDC defines depression as a score over 10.
▪ Otherwise, they are not depressed.
▪ The BDC score classified participants as depressed or not. 14 October 2025 12
Selected Features
14 October 2025 13
Data Visualization
14 October 2025 14
Simulation Result
14 October 2025 15
Continue
14 October 2025 16
Explainability Using SHAP
The Shapley additive explanations (SHAP) approach calculates the
significance score of individual aspects for a prediction.
The technique involves identifying a novel category of additive
feature significance measures and theoretically proving the
existence of a single solution with several desirable properties.
In [11], the authors used 15 features.
Our model uses 13 features and performs better.
SHAP analysis includes all features from [19] except SUCIDE and
EATDIS (Figure 1.6).
However, SHAP features match our 13 features.
SHAP analysis proves this.
14 October 2025 17
Continue
14 October 2025 18
Comparison
14 October 2025 19
Comparison (continue)
14 October 2025 20
Conclusion
Psychological Stress can be caused by several things.
The current study seeks to identify psychological stress’s factors.
To identify the demographic and stress factors that cause
depression several feature selection methods have been used.
During training, feature selection methods improve classifier
depression screening accuracy and efficiency.
Four well known machine learning classifiers were used to study
psychological stress.
The KNN classifier, which uses Genetic Search and Ranker feature
selection methods, can reliably predict psychological stress with
96.355% accuracy, according to this study.
14 October 2025 21
Future Research Direction
An expert system can be developed to identify
stress more efficiently by adopting the best machine
learning model.
14 October 2025 22
References
[1] Shumaila Aleem, Noor ul Huda, Rashid Amin, Samina Khalid, Sultan S Al- shamrani, and
Abdullah Alshehri. Machine learning algorithms for depression: diagnosis, insights, and
research directions. Electronics, 11(7):1111, 2022.
[2] Umut Arioz, Ur ıska Smrke, Nejc Plohl, and Izidor Mlakar. Scoping review on the
multimodal classification of depression and experimental study on existing multimodal
models. Diagnostics, 12(11):2683, 2022.
[3] Md Al-Mamun Biilah, M Raihan, Tamanna Akter, Nasif Alvi, Nusrat Jahan Bristy, and
Hasin Rehana. Human depression prediction using association rule mining technique. In
International Conference on Innovative Computing and Communications: Proceedings of
ICICC 2021, Volume 2, pages 223–237. Springer, 2022.
[4] Md Al-Mamun Billah, M Raihan, Nasif Alvi, Tamanna Akter, and Nusrat Jahan Bristy. A
data mining approach to identify the stress level based on different activities of human. In
2021 International Conference on Information and Communication Technology for
Sustainable Development (ICICT4SD), pages 31–34. IEEE, 2021.
[5] Ekaba Bisong and Ekaba Bisong. Google colaboratory. Building machine learning and
deep learning models on google cloud platform: a comprehensive guide for beginners,
pages 59–64, 2019
14 October 2025 23
References (continue)
[6] Salvatore Fara, Orlaith Hickey, Alexandra Georgescu, Stefano Goria, Emilia Molimpakis, and
Nicholas Cummins. Utilising bayesian networks to combine multimodal data and expert opinion for
the robust prediction of depression and its symptoms. arXiv preprint arXiv:2211.04924, 2022.
[7] Gaurav Kumar Gupta and Dilip Kumar Sharma. A review of overfitting solutions in smart
depression detection models. In 2022 9th International Conference on Computing for Sustainable
Global Development (INDIACom), pages 145–151. IEEE, 2022.
[8] Majharul Islam, Md Shohel Arman, Md Abu Bakkar Siddik, Selina Akter So- nia, Atqiah Rahman
Reesha, and Fahadul Islam. Mental stress analysis of bangladeshi job holders during covid-19
pandemic: A machine learning approach. In 2022 13th International Conference on Computing
Communication and Net- working Technologies (ICCCNT), pages 1–6. IEEE, 2022.
[9]Scott M. Lundberg and Su-In Lee. A unified approach to interpreting model predictions. In
Proceedings of the 31st International Conference on Neural In-formation Processing Systems,
NIPS’17, page 4768–4777, Red Hook, NY, USA, 2017. Curran Associates Inc.
[10]Amie J Moyes and Adrian J Hobbs. C-type natriuretic peptide: a multifaceted paracrine regulator
in the heart and vasculature. International journal of molec-ular sciences, 20(9):2281, 2019.
14 October 2025 24
References (continue)
[11]Katarzyna Nowicka-Sauer, Krzysztof Jarmoszewicz, Andrzej Molisz, Krzysztof Sobczak, Marta Sauer,
and Mariusz Topolski. Applying machine learning to construct a model of risk of depression in patients
following cardiac surgery with the use of the sf-12 survey. International Journal of Environmental
Research and Public Health, 20(6):4876, 2023.
[12]Sunday O Olatunji, Nawal Alsheikh, Lujain Alnajrani, Alhatoon Alanazy, Me-shael Almusairii, Salam
Alshammasi, Aisha Alansari, Rim Zaghdoud, Alaa Alah-madi, Mohammed Imran Basheer Ahmed, et al.
Comprehensible machine-learning-based models for the pre-emptive diagnosis of multiple sclerosis using
clinical data: A retrospective study in the eastern province of saudi arabia. In-ternational Journal of
Environmental Research and Public Health, 20(5):4261, 2023.
[13]M Raihan, Nasif Alvi, and Anupam Kumar Bairagi. Social behavior prediction using ensemble learning
algorithms. In 2022 13th International Conference on Computing Communication and Networking
Technologies (ICCCNT), pages 1–5. IEEE, 2022.
[14]M Raihan, Anupam Kumar Bairagi, and Shagoto Rahman. A machine learning based study to predict
depression with monitoring actigraph watch data. In 2021 12th International Conference on Computing
Communication and Networking Technologies (ICCCNT), pages 1–5. IEEE, 2021.
14 October 2025 25
References (continue)
[15]M Raihan, Md Tanvir Islam, Promila Ghosh, Md Mehedi Hassan, Jarif Huda An-gon, and Sajib Kabiraj.
Human behavior analysis using association rule mining techniques. In 2020 11th International conference
on computing, communication and networking technologies (ICCCNT), pages 1–5. IEEE, 2020.
[16]Neha Prerna Tigga and Shruti Garg. Efficacy of novel attention-based gated recurrent units
transformer for depression detection using electroencephalogram signals. Health Information Science and
Systems, 11(1):1–17, 2023.
[17]Ian H Witten and Eibe Frank. Data mining: practical machine learning tools and techniques with java
implementations. Acm Sigmod Record, 31(1):76–77, 2002.
[18]Yu Xin and Xiaohui Ren. Predicting depression among rural and urban disabled elderly in china using
a random forest classifier. BMC psychiatry, 22(1):1–11, 2022.
[19]Md Sabab Zulfiker, Nasrin Kabir, Al Amin Biswas, Tahmina Nazneen, and Mo-hammad Shorif Uddin. An
in-depth analysis of machine learning approaches to predict depression. Current research in behavioral
sciences, 2:100044, 2021.
14 October 2025 26
Thank You
14 October 2025 27