Arabic Clinical Mental Health Named Entity Recognition via Distant Supervision: Recurrent vs. Transformer-Based Architectures

Authors

• Ali A. Jalil, Cybersecurity Department, College of Science, University of AlKafeel, 54001, Najaf, IraqFollow
• Rabih Sbera, Computer Engineering Techniques Department, College of Engineering Technology, Ashur University, 10001, Baghdad, Iraq; Department of Computer and Automatic Control Engineering, College of Mechanical and Electrical Engineering, Latakia University, 41, Latakia, Syria

How to Cite This Article

Jalil, Ali A. and Sbera, Rabih (2026) "Arabic Clinical Mental Health Named Entity Recognition via Distant Supervision: Recurrent vs. Transformer-Based Architectures," Polytechnic Journal: Vol. 16: Iss. 2, Article 3.
DOI: https://doi.org/10.59341/2707-7799.1882

Document Type

Original Article

Abstract

Arabic Natural Language Processing (NLP) has made great strides, but the accurate extraction of mental state and physiological symptoms from patient-generated clinical consultations in dialectal Arabic, which is plagued with significant morphological complexity, colloquial noise and an extremely limited number of manually annotated clinical benchmarks, remains a significant challenge. In this work, we propose an end-to-end token-level symptom extraction system that is systematic and grounded in distant supervision paradigm and a strict gold standard evaluation with human verification to ensure real-world clinical generalization. A large weakly labelled corpus of 46,873 Arabic patient questions was created based on a clinical lexicon of 152 psychological and somatic symptoms, mapped to ICD-11 (Chapter 06) and DSM-5 criteria. To overcome the drawbacks of distant supervision and circular evaluation, a 500-consultation human-verified gold standard was developed, achieving an inter-annotator agreement of Cohen’s κ = 0.89 evaluated at the word token level across all 68,760 tokens. We have performed a thorough comparative study of four neural architectures: Bidirectional Gated Recurrent Unit (Bi-GRU), Pure Custom Transformer Encoder trained from scratch, Transformer with FastText augmentation, and AraBERT model fine-tuned using the data. We add a layer of 1D Convolutional Feature Representation (CFR) with k = 3 on all architectures to capture local subword/character-level morphological patterns and a Conditional Random Field (CRF) layer to enforce valid BIO syntax. AraBERT subword tokenization was aligned to word-level boundaries using first-subword pooling and loss masking (ignore_index = -100) to ensure a strictly standardized evaluation space across all models (712,287 original word tokens in the held-out test set, corresponding to 934,181 AraBERT subwords). Results & Contributions: On the human-verified gold standard, the fine-tuned AraBERT model demonstrates superior clinical semantic generalization, obtaining a Macro F1-score of 0.8735, outperforming Bi-GRU with 0.8214 by successfully resolving clinical negation and dialectal synonyms. Conversely, on the synthetic distant supervision test set, the Bi-GRU-CRF achieves a Macro F1-score of 0.9051 (compared to 0.8987 for AraBERT-CRF), providing empirical evidence of the effective rule replication and memorization of deterministic dictionary patterns by recurrent models. Crucially, complexity analysis reveals that Bi-GRU operates at a fraction of the computational footprint (6.31M parameters, 0.37 ms latency per sentence) compared to AraBERT (135.78M parameters, 11.77 ms latency), providing an ultra-efficient alternative for real-time edge clinical deployment. Conclusion: To the best of our knowledge, this work establishes the first comprehensive token-level Arabic clinical mental health NER benchmark. The integration of pre-trained language models with CRF sequential constraints, validated against a human-annotated benchmark, provides a robust methodology for Arabic clinical Named Entity Recognition (NER). Future directions include domain-adaptive clinical pretraining and active learning to further reduce manual annotation overhead.

Receive Date

26 June 2026

Revise Date

22 Aug 2026

Accept Date

22 Aug 2026

Publication Date

10-1-2026

References

[1] Guntuku SC, Yaden DB, Kern ML, Ungar LH, Eichstaedt JC. Detecting depression and mental illness on social media: an integrative review. Curr Opin Behav Sci 2017;18:43—9. https://doi.org/10.1016/j.cobeha.2017.07.007.

[2] De Choudhury M, Gamon M, Counts S, Horvitz E. Predicting depression via social media. In: Proceedings of the International AAAI Conference on Web and Social Media; 2013; 7(1)128-37. https://doi.org/10.1609/ icwsm.v7i1.14432.

[3] Coppersmith G, Dredze M, Harman C. Quantifying mental health signals on Twitter. In: Proceedings of the workshop on computational linguistics and clinical psychology: from linguistic signal to clinical reality, CLPsych Workshop ACL 2014, Baltimore, MD, USA. Stroudsburg, PA, USA: Association for Computational Linguistics (ACL); 2014. p. 51—60. https://doi.org/10.3115/v1/W14-3207.

[4] Nadkarni PM, Ohno-Machado L, Chapman WW. Natural language processing: an introduction. J Am Med Inform Assoc 2011;18(5):544—51. https://doi.org/10.1136/amiajnl- 2011-000464.

[5] Shaalan K. A survey of Arabic named entity recognition and classification. Comput Linguist 2014;40(2):469—510. https:// doi.org/10.1162/COLI_a_00178.

[6] Farghaly A, Shaalan K. Arabic natural language processing: Challenges and solutions. ACM Trans Asian Lang Inf Process 2009;8(4):1—22. https://doi.org/10.1145/1644879.1644881.

[7] Oudah M, Shaalan K. A Pipeline Arabic Named Entity Recognition using a Hybrid Approach. In: Proceedings of the 24th international conference on computational linguistics (COLING 2012). Mumbai, India: The COLING 2012 Organizing Committee/ACL; 2012. p. 2113—28. https:// aclanthology.org/C12-1132/.

[8] Benajiba Y, Rosso P. ANERsys: An Arabic named entity recognition system based on maximum entropy. In: International conference on intelligent text processing and computational linguistics, CICLing 2007/LNCS vol. 4394, Mexico City, Mexico. Berlin, Heidelberg, Germany: Springer Berlin Heidelberg; 2007. p. 143—53. https://doi.org/ 10.1007/978-3-540-70939-8_13.

[9] Alasmari A. A Scoping Review of Arabic Natural Language Processing for Mental Health. Healthcare 2025;13(9):963. https://doi.org/10.3390/healthcare13090963.

[10] Alhuzali H, Alasmari A, Alsaleh H. MentalQA: An Annotated Arabic Corpus for Questions and Answers of Mental Healthcare. IEEE Access 2024;12:101155—65. https://doi.org/ 10.1109/ACCESS.2024.3430068.

[11] Gannoune N, Madani A, Kissi M. Enhancing Arabic Biomedical Named Entity Recognition Using Transformer- Based Representations and CRF Sequence Labeling. Int J Adv Comput Sci Appl 2025;16(12):485—94. https://doi.org/ 10.14569/IJACSA.2025.0161256.

[12] Wali A, Almagrabi H, El-Feky S, Jokhdar M. Dawwen: An Arabic Mental Health Mobile App Based on Natural Language Processing. Int J Interact Mobile Technol 2025;19(4): 45—54. https://doi.org/10.3991/ijim.v19i04.51999.

[13] Elmajali S, Ahmad I. Toward Early Detection of Depression: Detecting Depression Symptoms in Arabic Tweets Using Pretrained Transformers. IEEE Access 2024;12:88134—45. https://doi.org/10.1109/ACCESS.2024.3417821.

[14] Zaghouani W, Shlkamy ES, Bessghaier M. From Posts to Pressure: An Arabic Dataset about Stress and Mental- Health Monitoring. In: Proceedings of the 2nd workshop on NLP for languages using Arabic script AbjadNLP Workshop @ EACL 2026, Rabat, Morocco. Stroudsburg, PA, USA: Association for Computational Linguistics (ACL); 2026. p. 234—43. https://doi.org/10.18653/v1/2026.abjadnlp-1.50.

[15] Jarrar M, Khalilia M, Ghanem S. Wojood: Nested Arabic Named Entity Corpus and Recognition using BERT. In: Proceedings of the thirteenth language resources and evaluation conference (LREC 2022, Marseille, France). Marseille, France: European Language Resources Association (ELRA); 2022. p. 3626—36. https://doi.org/10.48550/ arXiv.2205.09651.

[16] Kocaman V, Mellah Y, Haq HU, Talby D. Automated De- Identification of Arabic Medical Records. In: Proceedings of the first Arabic natural language processing conference (ArabicNLP 2023, Singapore). Stroudsburg, PA, USA: Association for Computational Linguistics (ACL); 2023. p. 33—43. https://doi.org/10.18653/v1/2023.arabicnlp-1.4.

[17] Oudah M, Shaalan K. Studying the impact of language-independent and language-specific features on hybrid Arabic Person name recognition. Language Resour Eval 2017;51(2): 351—78. https://doi.org/10.1007/s10579-016-9376-1.

[18] Alshammari N, Alanazi S. An Arabic Dataset for Disease Named Entity Recognition with Multi-Annotation Schemes. Data 2020;5(3):60. https://doi.org/10.3390/data5030060.

[19] Al-Twairesh N, et al. AHD: Arabic Healthcare Dataset for Medical Applications. Mendeley Data 2023;V5. https://doi. org/10.17632/mgj29ndgrk.5.

[20] Al-Nassan MA, Al-Ayyoub M, Jararweh Y, Benkhelifa E. AltibbiVec: A Word Embedding Model for the Medical Arabic Domain. Procedia Comput Sci 2021;182:10—7. https:// doi.org/10.1016/j.procs.2021.02.002.

[21] World Health Organization. International classification of diseases. 11th Revision. Geneva: WHO; 2019 (ICD-11). https://icd.who.int/en.

[22] American Psychiatric Association. Diagnostic and statistical manual of mental disorders. fifth ed. Arlington, VA: American Psychiatric Publishing; 2013. https://doi.org/10. 1176/appi.books.9780890425596 (DSM-5).

[23] Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med 2001;16(9):606—13. https://doi.org/10.1046/j.1525-1497.2001. 016009606.x.

[24] Spitzer RL, Kroenke K, Williams JB, L owe B. A brief measure for assessing generalized anxiety disorder: the GAD-7. Arch Intern Med 2006;166(10):1092—7. https://doi.org/10. 1001/archinte.166.10.1092.

[25] Gridach M. Character-level neural network for biomedical named entity recognition. J Biomed Inf 2017;70:85—91. https://doi.org/10.1016/j.jbi.2017.05.002.

[26] Chung J, Gulcehre C, Cho K, Bengio Y. Empirical evaluation of gated recurrent neural networks on sequence modeling. NeurIPS Gate Recurr Netw Workshop 2014:1—9. https://doi. org/10.48550/arXiv.1412.3555.

[27] Lample G, Ballesteros M, Subramanian S, Kawakami K, Dyer C. Neural architectures for named entity recognition. In: Proceedings of the 2016 conference of the north American chapter of the association for computational linguistics (NAACL-HLT 2016, San Diego, CA, USA). Stroudsburg, PA, USA: Association for Computational Linguistics (ACL); 2016. p. 260—70. https://doi.org/10.18653/v1/N16-1030.

[28] Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, et al. Attention is all you need. Adv Neural Inf Process Syst 2017;30:5998—6008. https://doi.org/10.48550/ arXiv.1706.03762.

[29] Ba JL, Kiros JR, Hinton GE. Layer normalization. arXiv preprint arXiv:160706450 2016. https://doi.org/10.48550/ arXiv.1607.06450.

[30] Antoun W, Baly F, Hajj H. AraBERT: Transformer-based model for Arabic language understanding. In: Proceedings of the 4th workshop on open-source Arabic Corpora and processing tools (OSACT 4 @ LREC 2020, Marseille, France). Marseille, France: European Language Resources Association (ELRA); 2020. p. 9—15. https://doi.org/10.48550/arXiv. 2003.00104.

[31] Abdul-Mageed M, Elmadany A, Nagoudi EMB. ARBERT & MARBERT: Deep Bidirectional Transformers for Arabic. In: Proceedings of the 59th annual meeting of the association for computational linguistics (ACL-IJCNLP 2021, Online). Stroudsburg, PA, USA: Association for Computational Linguistics (ACL); 2021. p. 7088—105. https://doi.org/10.18653/ v1/2021.acl-long.551.

[32] Al-Twairesh N. The evolution of language models applied to emotion analysis of Arabic tweets. Information 2021;12(2): 84. https://doi.org/10.3390/info12020084.

[33] Bojanowski P, Grave E, Joulin A, Mikolov T. Enriching word vectors with subword information. Trans Assoc Comput Linguist 2017;5:135—46. https://doi.org/10.1162/ tacl_a_00051.

[34] Devlin J, Chang MW, Lee K, Toutanova K. BERT: Pre- training of deep bidirectional transformers for language understanding. In: Proceedings of the 2019 conference of the north American chapter of the association for computational linguistics (NAACL-HLT 2019, Minneapolis, MN, USA). Stroudsburg, PA, USA: Association for Computational Linguistics (ACL); 2019. p. 4171—86. https://doi.org/10. 18653/v1/N19-1423.

[35] Lin TY, Goyal P, Girshick R, He K, Dollar P. Focal loss for dense object detection. In: Proceedings of the IEEE international conference on computer vision (IEEE ICCV 2017, Venice, Italy). Piscataway, NJ, USA: IEEE (Institute of Electrical and Electronics Engineers); 2017. p. 2980—8. https://doi.org/10.1109/ICCV.2017.324.

[36] Yadav V, Bethard S. A survey on recent advances in named entity recognition from deep learning models. Proceedings of the 27th International Conference on Computational Linguistics (COLING). 2018. p. 2145—58.

[37] Uzuner  O, Solti I, Xia F. 2010 i2b2/VA challenge on concepts, assertions, and relations in clinical text. J Am Med Inform Assoc 2011;18(5):552—6. https://doi.org/10.1136/ amiajnl-2011-000203.