Natural Language Processing Applied to Psychiatric Clinical Notes: Scoping Review

Clinical Snapshot

Key Findings

• P Value: Not applicable — no inferential statistical testing performed

• Effect Size: Rule-based methods: n=36 (35.6%); Hybrid approaches: n=34 (33.7%); Deep learning: n=15 (14.9%); Traditional ML: n=10 (9.9%); LLM-based: n=6 (5.9%). Hybrid and rule-based methods collectively account for 69.3% of included studies.

• Primary Outcome: Characterisation and frequency distribution of NLP methodologies applied to psychiatric clinical notes across 101 eligible studies (2021–2025)

• Nnt Or Sensitivity: Not applicable — no pooled diagnostic accuracy, NNT, or hazard ratio calculable from this scoping review. Individual study performance metrics (e.g., F1 scores, AUC) are referenced narratively but not pooled.

• Confidence Interval: Not applicable — scoping review with descriptive synthesis only; no confidence intervals reported

Clinical Application

Direct clinical implementation of specific NLP tools is not recommended based on this scoping review alone. The review serves as a research roadmap rather than a deployment guide. Feasibility of NLP implementation in psychiatric settings depends on EHR system compatibility, institutional data governance frameworks, clinician workflow integration, and availability of annotated training data. The dominance of hybrid and rule-based methods suggests that fully automated NLP pipelines for psychiatric notes remain technically challenging. In Australia, the Mental Health Act frameworks across states and territories generate substantial structured and unstructured clinical documentation. The My Health Record system and state-based EHR platforms (e.g., NSW Health's eMR, Queensland Health's ieMR) contain psychiatric clinical notes that could theoretically benefit from NLP-assisted analysis. However, no Australian-specific studies are identifiable from this review's abstract. The RACGP and the Royal Australian and New Zealand College of Psychiatrists (RANZCP) have not issued specific guidance on NLP deployment in psychiatric practice. TGA regulatory pathways for software as a medical device (SaMD) would apply to any clinically deployed NLP tool. Privacy considerations under the Australian Privacy Act 1988 and the My Health Records Act 2012 are particularly salient given the sensitive nature of psychiatric data. PBS implications are not directly relevant to NLP methodology reviews. Mental health clinicians, health informaticians, and clinical researchers working with psychiatric EHR data across inpatient, outpatient, and community mental health settings. Most directly applicable to institutions with structured EHR systems containing free-text psychiatric notes.

Abstract

BACKGROUND: Psychiatric clinical notes in electronic health records (EHRs) provide rich longitudinal information that can support clinical decision-making. Using historical medical data can enable earlier identification of mental illness, better characterization of disease trajectories, and more personalized treatment planning. Natural language processing (NLP) transforms these unstructured notes into analyzable representations for research and care. OBJECTIVE: This study aims to systematically summarize NLP methodologies for psychiatric clinical notes, compare major modeling paradigms and application areas, and highlight emerging large language model (LLM) trends, key challenges, and future research directions. METHODS: Following the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) guidelines, a literature search was conducted for articles on NLP methods based on psychiatric clinical notes published from January 2021 to December 2025 in Ovid MEDLINE, Ovid EMBASE, PubMed, Scopus, Web of Science, the ACM Digital Library, and ScienceDirect. This scoping review analyzed NLP methods applied to psychiatric clinical notes, focusing on major trends, identifying suitable features for traditional machine learning (ML)-based models, applications of pretrained language models (PLMs), and key challenges. Approaches were categorized as rule-based, traditional ML, hybrid, deep learning (DL), and LLM-based methods across information extraction and text classification tasks. RESULTS: In total, 101 studies were eligible for inclusion. Rule-based methods (n=36) and hybrid approaches (n=34) remained the most widely used techniques, largely favored for their interpretability in handling nuanced, subjective clinical notes. These were followed by DL (n=15), traditional ML (n=10), and LLM-based approaches (n=6). Traditional ML studies relied heavily on engineered features, which could be grouped into 5 broad categories: domain knowledge features, lexical and statistical features, vector-based semantic features, emotion-related features, and temporal features. PLMs improved performance mainly through domain adaptation and task-specific fine-tuning, enhancing the handling of psychiatric language, medical terminology, and clinical note structure. LLM-based studies, although still limited in number, indicated a growing shift toward generative and reasoning-based applications. CONCLUSIONS: Hybrid NLP approaches remain dominant, combining domain rules with ML for extraction and classification. DL approaches continue to advance, with domain adaptation supporting medical terminology and clinical semantics. LLMs may further automate complex workflows via zero-shot capabilities and reasoning, alongside growing interest in temporal modeling and multimodal integration. Key future needs include improved generalizability across institutions, privacy protection, and careful attention to ethical implications in clinical deployment.

References

1. 1.Rao, S., Chen, X., Deng, G., Xie, J., Jiang, T., Li, T., Zhang, Y., & Jiang, H. (2026). Natural language processing applied to psychiatric clinical notes: Scoping review. JMIR Medical Informatics. https://doi.org/10.2196/91249