Analyzing multidimensional symptom cluster using unstructured electronic health records: A scoping review protocol

Background: Patients with chronic or severe illnesses frequently experience multiple interrelated symptoms that co-occur as symptom clusters, exacerbating functional impairment, treatment non-adherence, and reduced quality of life. Electronic health records (EHRs), particularly unstructured data such as clinical notes and patient narratives, will offer unique opportunities to uncover the complexity and temporal dynamics of symptom clusters. Although natural language processing (NLP) and machine learning techniques are increasingly used to extract symptom information from EHRs, the methodological landscape of how symptom clusters are derived from EHR data remains underexplored. Objective: This scoping review will aim to systematically identify how symptom clusters have been derived from EHR data and to evaluate the analytic methods used in this emerging area of research. Specifically, the review will address two guiding questions: (1) How have unstructured EHR data (e.g., clinical notes, patient narratives) been used to identify symptom clusters? and (2) What methodological approaches have been employed to derive symptom clusters from EHR data? Methods: This scoping review will follow the JBI Manual for Evidence Synthesis and PRISMA-ScR guidelines. Using the PCC framework, we will include studies of hospitalized adults with chronic or serious illnesses that examine symptom clusters using EHR data. A medical librarian will assist in developing a comprehensive search across MEDLINE, Embase, CINAHL, APA PsycINFO, and Web of Science. Two reviewers will independently screen studies and extract data using a structured charting form. Findings will be synthesized narratively and visually to summarize study characteristics, EHR data sources, symptom measures, analytic methods, and key results. Discussion: This review will synthesize how EHR data—especially unstructured clinical text—have been used to identify symptom clusters. By mapping methodological approaches and highlighting strengths, limitations, and gaps, it will guide the development of more standardized and clinically meaningful methods for EHR-based symptom cluster research. The results will support researchers, clinicians, and informaticians in advancing mechanism-informed interventions and improving symptom management for patients with complex chronic conditions.