Autoinflammatory & rare systemic diseases
Machine-learning algorithms for predicting colchicine resistance in Familial Mediterranean Fever
Authors' conclusion
Our findings suggest that AI-based algorithms, particularly DL models, show significant potential in predicting colchicine resistance in adult patients with FMF.
Study context
- Population
- Adults with FMF diagnosed by the Tel Hashomer criteria, genetically confirmed, with at least 1 year of follow-up
- Participants
- 965 adults
- Exposure
- Logistic regression model vs fully connected neural network (deep-learning (DL)) for predicting colchicine resistance
- Outcome
- Prediction of colchicine resistance: AUC of the ROC curve
Methodological detailsmachine-learning prediction model development, 80:20 training/testing split, logistic regression, fully connected neural network
- Machine-learning prediction model development
- 80:20 training/testing split
- Logistic regression
- Fully connected neural network
Main finding
As reported in the abstract
AUC for predicting colchicine resistance
| Group | AUC |
|---|---|
| Logistic regression | 0.79 |
| DL | 0.79 |
AUC
Logistic regression0.79
DL0.79
Observational data: associations, not effects.
Further findings
Features differing between colchicine-resistant and non-resistant patients
- Homozygous mutation type
- P = 0.0005
- Recurrent arthritis
- P = 0.0033
- Chronic arthralgia
- P = 0.0286
- Age of diagnosis
- P = 0.0022
- Frequency of attacks
- P = 0.0436
- Statistically significant differences
Original publication
Machine-learning algorithms for predicting colchicine resistance in Familial Mediterranean Fever. Rheumatology (Oxford). Published 17 February 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →