Evaluating the Accuracy of the Frysian Questionnaire for Differentiation of Musculoskeletal Complaints for Triage of Musculoskeletal Diseases: Algorithm Development and Validation Study
Bibliographic record
Abstract
Background: Inflammatory rheumatic diseases (IRDs) affect 5% of the general population, whereas 35% of the population experiences musculoskeletal concerns. IRDs cause early disability, reduced life expectancy, and considerable health care costs. Early diagnosis is essential to prevent long-term damage. Similarly important is the early identification of patients with musculoskeletal concerns without IRDs to prevent unnecessary health care expenses. Of the population referred to the rheumatologist, 60% have noninflammatory musculoskeletal concerns, whereas only 20% of patients with an IRD see a rheumatologist within 3 months of symptom onset. The need for digital predictive (triage) tools for rheumatic and musculoskeletal diseases led to the development of the Frysian Questionnaire for Differentiation of Musculoskeletal Complaints (FRYQ). Objective: This study aimed to assess whether the FRYQ can distinguish IRD from noninflammatory musculoskeletal concerns in general, and rheumatoid arthritis and fibromyalgia specifically, in newly referred patients. Methods: The FRYQ is an 87-item tool (20 open-ended and 67 closed-ended questions) used to triage new rheumatology patients at Frisius Medical Center in the Netherlands. We analyzed data from 2 sources: dataset A with 728 outpatient clinic patients and dataset B with 373 patients from the Joint Pain Assessment Scoring Tool study. We built a classifier using Extreme Gradient Boosting to distinguish inflammatory from noninflammatory conditions based on closed-ended questions. Using elastic net regularization, we identified the most informative questions. We evaluated classification using receiver operating characteristic curve analysis and assessed feature importance through Shapley Additive Explanation analysis. To test generalizability, we replicated our analysis on dataset B. Finally, we examined whether the questions of the FRYQ could be used to identify specific conditions beyond the general categories of IRD and non-IRD, specifically for detecting fibromyalgia and rheumatoid arthritis. Results: Feature selection reduced the questionnaire from 67 to 28 items while maintaining discriminative power. After initial development, the model achieved an area under the receiver operating characteristic curve (AUC-ROC) of 0.72 (95% CI 0.67-0.78) for distinguishing inflammatory from noninflammatory conditions in an external validation set. Using a probability threshold of 0.30, the model achieved 71% sensitivity and 56% specificity on external validation. The FRYQ demonstrated stronger performance in identifying specific diagnoses such as fibromyalgia (AUC-ROC=0.81) and rheumatoid arthritis (AUC-ROC=0.77). Key discriminating features included symptom duration, pain response to movement, and anti-inflammatory medication effectiveness. Conclusions: The FRYQ effectively distinguishes inflammatory from noninflammatory rheumatic conditions before specialist consultation and shows particular strength in identifying fibromyalgia and rheumatoid arthritis. This tool could improve rheumatology triage by prioritizing referrals with high likelihood of IRD for early rheumatologist evaluation while directing other patients to appropriate alternative resources. Prospective studies are needed to determine the FRYQ's impact on clinical outcomes and health care efficiency.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.018 | 0.027 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.003 | 0.002 |
| Bibliometrics | 0.003 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.003 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".