{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,7]],"date-time":"2026-10-07T06:49:41Z","timestamp":1791355781567,"version":"4.3.1"},"reference-count":31,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,10,7]],"date-time":"2026-10-07T00:00:00Z","timestamp":1791331200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Artif. Intell."],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Depression detection models for Arabic social-media text often achieve high predictive accuracy but operate as black boxes, limiting their acceptability in clinical and public-health contexts. This study aimed to develop and evaluate an explainable machine learning framework capable of identifying clinically meaningful linguistic markers of depression in Arabic tweets.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>\n                      We used the ArabMindGuard dataset (3,079 manually annotated tweets; 21.4% Depressed). Tweets were processed through a standard Arabic NLP pipeline and represented using TF-IDF vectorization (unigrams and bigrams). We compared six classification pipelines (Logistic Regression, Random Forest with three imbalance-handling strategies, SVM, and XGBoost) using 5-fold stratified cross-validation with TF-IDF and resampling strictly inside each fold. The primary model, Random Forest with SMOTE, was selected for its compatibility with exact TreeSHAP computations. To assess generalizability, we externally validated the model on the independent Nassar corpus (\n                      <jats:italic>N<\/jats:italic>\n                      = 10,000). TreeSHAP was applied for interpretability, and a systematic error analysis was conducted on all false negatives and false positives.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>The RF + SMOTE model achieved an AUC-ROC of 0.80 on the held-out test set and a weighted F1-score of 0.79, with SMOTE improving depressed recall from 0.38 to 0.57 (CV recall = 0.59 \u00b1 0.05). SHAP analysis identified \u0645\u0627\u0641\u064a\u0646\u064a (lack of energy), \u0646\u0641\u0633\u064a\u062a\u064a (psychological distress), \u062e\u0645\u0648\u0644 (lethargy), \u0636\u064a\u0642\u0647 (distress) and \u0637\u0627\u0642\u0629 \u0645\u0627\u0641\u064a\u0646\u064a (I have no energy) as the most influential predictors, aligning with DSM-5 symptom domains. External validation on Nassar yielded an AUC of 0.752, with 80% (16\/20) of top SHAP features overlapping across datasets. Systematic error analysis of 58 false negatives and 67 false positives revealed that false negatives consistently lacked these prototypical markers, confirming the model's interpretable decision logic.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>By combining a transparent classifier with SHAP-based interpretation and cross-dataset validation, this study provides a replicable framework for surfacing clinically meaningful linguistic markers of depression in Arabic text. The approach is intended to support, rather than replace, clinical judgment, and offers a methodological template for transparent digital mental health screening tools in under-resourced languages.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/frai.2026.1917139","type":"journal-article","created":{"date-parts":[[2026,10,7]],"date-time":"2026-10-07T05:52:16Z","timestamp":1791352336000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Explainable detection of depression in Arabic tweets: identifying clinically meaningful linguistic markers using SHAP"],"prefix":"10.3389","volume":"9","author":[{"given":"Erfan","family":"Salehi","sequence":"first","affiliation":[{"name":"Department of Clinical Psychology, Faculty of Medicine, Hamedan Branch, Islamic Azad University","place":["Hamedan, Iran"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Maryam","family":"Raissi","sequence":"additional","affiliation":[{"name":"Department of Psychology, Science and Research Branch, Islamic Azad University","place":["Tehran, Iran"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Amirhossein","family":"Rasouli","sequence":"additional","affiliation":[{"name":"Student Research Committee, Department of Clinical Psychology, School of Medicine, Shahid Beheshti University of Medical Sciences","place":["Tehran, Iran"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wajdi","family":"Zaghouani","sequence":"additional","affiliation":[{"name":"Communication Program, Northwestern University in Qatar, Education City","place":["Doha, Qatar"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,10,7]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"12729","DOI":"10.1007\/s13369-024-08767-3","article-title":"Detecting suicidality in Arabic tweets using machine learning and deep learning techniques","volume":"49","author":"Abdulsalam","year":"2024","journal-title":"Arab. J. Sci. Eng"},{"key":"B2","unstructured":"Machine learning for detecting depression in Arabic tweets\n          \n          345\n          352\n          \n            \n              Alaskar\n              H.\n            \n            \n              Ykhlef\n              M.\n            \n          \n          Int. J. Adv. Comput. Sci. Appl\n          13\n          2022"},{"key":"B3","doi-asserted-by":"publisher","first-page":"963","DOI":"10.3390\/healthcare13090963","article-title":"scoping review of Arabic natural language processing for mental health","volume":"13","author":"Alasmari","year":"2025","journal-title":"Healthcare"},{"key":"B4","doi-asserted-by":"publisher","DOI":"10.1109\/ICCIT63348.2025.10989433","article-title":"\u201cX as a mirror for mental health: the role of AI in analyzing tweets for depression,\u201d","author":"Al-Atwi","year":"2025","journal-title":"2025 4th International Conference on Computing and Information Technology (ICCIT) University of Tabuk"},{"key":"B5","doi-asserted-by":"publisher","first-page":"e90321","DOI":"10.7759\/cureus.90321","article-title":"Prevalence of depression among urban and rural residents of Saudi Arabia compared with other Gulf Cooperation Council countries: a systematic review","volume":"17","author":"Alhomaid","year":"2025","journal-title":"Cureus"},{"key":"B6","doi-asserted-by":"publisher","first-page":"120","DOI":"10.1007\/s44163-026-00842-y","article-title":"Machine learning methods for detecting depression in Arabic tweets: a comprehensive performance analysis with enhanced evaluation metrics","volume":"6","author":"Aljameel","year":"2026","journal-title":"Discov. Appl. Sci"},{"key":"B7","doi-asserted-by":"publisher","first-page":"257","DOI":"10.1016\/j.procs.2019.12.107","article-title":"Detecting Arabic depressed users from Twitter data","volume":"163","author":"Almouzini","year":"2019","journal-title":"Procedia Comput. Sci"},{"key":"B8","doi-asserted-by":"publisher","DOI":"10.1109\/NCCC57165.2022.10067346","article-title":"\u201cDepression detection through identifying depressive Arabic tweets from Saudi Arabia: machine learning approach,\u201d","author":"Al-Musallam","year":"2022","journal-title":"2022 Fifth National Conference of Saudi Computers Colleges (NCCC)"},{"key":"B9","doi-asserted-by":"publisher","first-page":"1825","DOI":"10.1587\/transinf.2020EDP7023","article-title":"Machine learning-based approach for depression detection in twitter using content and activity features","volume":"103","author":"AlSagri","year":"2020","journal-title":"IEICE Trans. Inf. Syst"},{"key":"B10","doi-asserted-by":"publisher","first-page":"1441695","DOI":"10.3389\/fpubh.2024.1441695","article-title":"Prevalence of depression and anxiety among university students in Jeddah, Saudi Arabia: exploring sociodemographic and associated factors","volume":"12","author":"Alzahrani","year":"2024","journal-title":"Front. Public Health"},{"key":"B11","volume-title":"Diagnostic and Statistical Manual of Mental Disorders, 5th Edn","year":"2013"},{"key":"B12","doi-asserted-by":"publisher","first-page":"1609580","DOI":"10.3389\/fpsyt.2025.1609580","article-title":"How many people die by suicide each year? Not 727,000: a systematic review and meta-analysis of suicide underreporting across 71 countries over 122 years","volume":"16","author":"Daoud","year":"2025","journal-title":"Front. Psychiatry"},{"key":"B13","doi-asserted-by":"publisher","first-page":"88134","DOI":"10.1109\/ACCESS.2024.3417821","article-title":"Toward early detection of depression: detecting depression symptoms in Arabic tweets using pretrained transformers","volume":"12","author":"Elmajali","year":"2024","journal-title":"IEEE Access"},{"key":"B14","doi-asserted-by":"publisher","DOI":"10.1109\/ICICIS52592.2021.9694178","article-title":"\u201cCairoDep: detecting depression in Arabic posts using BERT transformers,\u201d","author":"El-Ramly","year":"2021","journal-title":"2021 Tenth International Conference on Intelligent Computing and Information Systems (ICICIS)"},{"key":"B15","doi-asserted-by":"publisher","first-page":"1517","DOI":"10.1001\/jama.2017.3826","article-title":"Depression is the leading cause of disability around the world","volume":"317","author":"Friedrich","year":"2017","journal-title":"JAMA"},{"key":"B16","doi-asserted-by":"crossref","DOI":"10.1007\/978-3-031-02139-8","volume-title":"Introduction to Arabic Natural Language Processing","author":"Habash","year":"2010"},{"key":"B17","article-title":"Enhanced suicidal ideation detection from social media using a CNN-BiLSTM hybrid model with SHAP-based interpretability","author":"Hasan","year":"2025","journal-title":"arXiv"},{"key":"B18","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.wanlp-1.28","article-title":"\u201cAradepsu: detecting depression and suicidal ideation in Arabic tweets using transformers,\u201d","author":"Hassib","year":"2022","journal-title":"Proceedings of the Seventh Arabic Natural Language Processing Workshop (WANLP)"},{"key":"B19","doi-asserted-by":"publisher","first-page":"102716","DOI":"10.1016\/j.artmed.2023.102716","article-title":"Depression detection for twitter users using sentiment analysis in English and Arabic tweets","volume":"147","author":"Helmy","year":"2024","journal-title":"Artif. Intell. Med"},{"key":"B20","unstructured":"Cultural variations in the clinical presentation of depression and anxiety: implications for diagnosis and treatment\n          \n          \n            \n              Kirmayer\n              L. J.\n            \n          \n          J. Clin. Psychiatry\n          62\n          2001"},{"key":"B21","doi-asserted-by":"publisher","first-page":"56","DOI":"10.1038\/s42256-019-0138-9","article-title":"From local explanations to global understanding with explainable AI for trees","volume":"2","author":"Lundberg","year":"2020","journal-title":"Nat. Mach. Intell"},{"key":"B22","article-title":"\u201cA unified approach to interpreting model predictions,\u201d","volume-title":"Advances in Neural Information Processing Systems 30 (NeurIPS)","author":"Lundberg","year":"2017"},{"key":"B23","doi-asserted-by":"publisher","first-page":"107812","DOI":"10.1016\/j.dib.2022.107999","article-title":"Modern Standard Arabic mood changing and depression dataset","volume":"40","author":"Maghraby","year":"2022","journal-title":"Data Brief"},{"key":"B24","doi-asserted-by":"publisher","first-page":"101186","DOI":"10.1016\/j.cogsys.2023.101186","article-title":"XAI Transformer based approach for interpreting depressed and suicidal user behavior on online social networks","volume":"83","author":"Malhotra","year":"2023","journal-title":"Cogn. Syst. Res"},{"key":"B25","doi-asserted-by":"publisher","first-page":"1412","DOI":"10.3390\/diagnostics15111412","article-title":"Explainable machine learning in the prediction of depression","volume":"15","author":"Mimikou","year":"2025","journal-title":"Diagnostics"},{"key":"B26","doi-asserted-by":"publisher","first-page":"237","DOI":"10.32604\/cmc.2022.022508","article-title":"Detecting depression in Arabic tweets using machine learning and NLP techniques","volume":"22","author":"Musleh","year":"2022","journal-title":"Int. J. Comput. Sci. Netw. Secur"},{"key":"B27","doi-asserted-by":"publisher","first-page":"7","DOI":"10.1186\/s43088-024-00592-9","article-title":"Recognition model for major depressive disorder in Arabic user-generated content","volume":"14","author":"Rabie","year":"2025","journal-title":"Beni Suef. Univ. J. Basic Appl. Sci"},{"key":"B28","doi-asserted-by":"publisher","first-page":"13006","DOI":"10.1038\/s41598-017-12961-9","article-title":"Forecasting the onset and course of mental illness with Twitter data","volume":"7","author":"Reece","year":"2017","journal-title":"Sci. Rep"},{"key":"B29","doi-asserted-by":"publisher","first-page":"543","DOI":"10.1177\/24056456251385719","article-title":"Generative contextual insights for social media mental health detection with Shapley Additive Explanations (SHAP)-based interpretation","volume":"23","author":"Tanaka","year":"2025","journal-title":"Web Intell"},{"key":"B30","doi-asserted-by":"publisher","first-page":"1476","DOI":"10.3390\/bs15111476","article-title":"Systematic review and meta-analysis of explainable machine learning models for clinical depression detection","volume":"15","author":"Trelles","year":"2025","journal-title":"Behav. Sci."},{"key":"B31","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2026.abjadnlp-1.50","article-title":"\u201cFrom posts to pressure: an Arabic dataset about stress and mental-health monitoring,\u201d","author":"Zaghouani","year":"2026","journal-title":"Proceedings of the 2nd Workshop on NLP for Languages Using Arabic Script"}],"container-title":["Frontiers in Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1917139\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,7]],"date-time":"2026-10-07T05:52:18Z","timestamp":1791352338000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1917139\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,10,7]]},"references-count":31,"alternative-id":["10.3389\/frai.2026.1917139"],"URL":"https:\/\/doi.org\/10.3389\/frai.2026.1917139","relation":{},"ISSN":["2624-8212"],"issn-type":[{"value":"2624-8212","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,10,7]]},"article-number":"1917139"}}