Objective:

To develop an algorithm based on a CHAID decision tree for the structured risk stratification of rare rheumatic diseases in a tertiary referral setting with unclear systemic complaints.

Approach:

• Statistical Analysis: Group differences were analyzed descriptively using inferential statistics, including t-tests and chi-square tests, and a CHAID decision tree model was constructed for risk stratification.

Key Findings:

• 52.0% of patients received a confirmed rheumatologic diagnosis.
• Fatigue and generalized pain were the most prevalent symptoms across diagnostic groups.
• The RHEUMA group had significantly higher scores in selected symptom domains and an immunoserological laboratory index.
• The CHAID-based decision model achieved an apparent classification of 81.5% (AUC = 0.893), outperforming logistic regression (76.3%, AUC = 0.823).

Interpretation:

The decision tree model provides a framework for structured risk stratification in a tertiary referral population, illustrating the use of symptom aggregation to support clinical reasoning.

Limitations:

• Monocentric, retrospective, and exploratory study design, which may limit the generalizability of the findings.
• Findings should be interpreted as hypothesis-generating, necessitating further research.
• External validation in independent cohorts is required for assessing model generalizability.

Conclusion:

The approach illustrates the potential of structured clinical reasoning in complex multisystem presentations, while acknowledging the need for further validation.