Novel multiclass classification machine learning approach for the early-stage classification of systemic autoimmune rheumatic diseases
Authors' conclusion
This pioneering multiclass ML model, using basic laboratory indicators, enhances clinical feasibility and demonstrates promising potential for SARDs classification.
Study context
- Population
- Systemic autoimmune rheumatic diseases (SARDs) patients categorised into SLE, Sjögren's syndrome (SS) and inflammatory myositis (IM); model built on nine accessible laboratory indicators
- Participants
- 925 patients2 cohorts
- Comparison
- Random forest (RF) tripartite classification model distinguishing SLE, IM and SS
- Outcome
- RF model classification performance (cohort 1)
Methodological detailsmulticlass machine learning, external validation
- Multiclass machine learning
- External validation
Performance
As reported in the abstract
| Test or model | AUC | Sensitivity | Specificity | PPV | NPV | Accuracy | Precision | Recall | F1 score |
|---|---|---|---|---|---|---|---|---|---|
| SLE | 0.953 | 0.890 | 0.910 | 0.922 | 0.854 | 0.892 | 0.930 | 0.985 | 0.957 |
| IM | 0.903 | 0.868 | 0.836 | 0.727 | 0.915 | 0.869 | 0.793 | 0.920 | 0.852 |
| SS | 0.836 | 0.795 | 0.748 | 0.663 | 0.879 | 0.857 | 0.950 | 0.679 | 0.792 |
AUC
SLE0.953
IM0.903
SS0.836
Sensitivity
SLE0.890
IM0.868
SS0.795
Specificity
SLE0.910
IM0.836
SS0.748
PPV
SLE0.922
IM0.727
SS0.663
NPV
SLE0.854
IM0.915
SS0.879
Accuracy
SLE0.892
IM0.869
SS0.857
Precision
SLE0.930
IM0.793
SS0.950
Recall
SLE0.985
IM0.920
SS0.679
F1 score
SLE0.957
IM0.852
SS0.792
Further findings
External validation (cohort 2), overall accuracy
- Overall accuracy
- 87.3%
SHAP analysis
- Significant contributions from antibody profiles
Original publication
Novel multiclass classification machine learning approach for the early-stage classification of systemic autoimmune rheumatic diseases. Lupus Sci Med. Published 31 January 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →