Clinical decision support systems for the management of low back pain: A systematic review

Background and aim: Low back pain (LBP) is one of the most prevalent musculoskeletal disorders and a major contributor to years lived with disability worldwide. Despite international guidelines promoting early conservative management and rational imaging use, clinical practice remains highly variable. Clinical Decision Support Systems (CDSS), particularly those integrating artificial intelligence (AI), have been developed to support diagnostic accuracy and standardize care. This review aims to synthesize current evidence on the performance, applications, and clinical integration of CDSS in the management of LBP.

Methods: A systematic review was conducted according to PRISMA 2020 guidelines. Searches were performed in PubMed, Scopus, ProQuest, and PsycINFO up to July 2025, including original studies evaluating CDSS for diagnosis or treatment planning in adults with LBP. Data extraction covered study design, CDSS type, data sources, performance metrics, and clinical outcomes. Risk of bias was assessed using QUADAS-2/QUADAS-AI, RoB 2, and ROBINS-I tools as appropriate. Descriptive statistics were computed for accuracy, sensitivity, specificity, and area under the curve (AUC).

Results: Nineteen studies met the inclusion criteria. Mean diagnostic accuracy was 0.911 (median 0.916), with corresponding mean sensitivity, specificity, and AUC of 0.865, 0.896, and 0.830, respectively. AI-based and hybrid systems performed comparably to rule-based models, while imaging optimization studies showed reductions of approximately 10% in unnecessary imaging and 15% in MRI utilization.

Conclusions: CDSS demonstrate high diagnostic performance and potential to improve guideline adherence and resource efficiency in LBP care. Broader implementation requires evaluation of long-term patient outcomes, cost-effectiveness, and real-world integration within electronic health records.