Detecting inflammatory arthritis in hand smartphone photographs: development and validation of a computer vision model in clinical settings
Authors' conclusion
A computer vision model trained on standardized smartphone photographs can detect hand synovitis in routine clinical populations including in those with deformities.
Study context
- Population
- Patients attending rheumatology clinics in India; standardized smartphone hand photographs; ground truth specialist-detected synovitis2,296 hand photos
- Participants
- 1,112 patients
- Comparison
- ConvNeXt V2 vs EfficientNetV2 vs their weighted ensemble, against specialist-detected synovitis
- Outcome
- AUROC for detecting hand synovitis
Methodological detailsdevelopment and validation of a computer vision model, patient-level partition into training (70%), validation (15%) and test (15%) sets, patient-level bootstrap 95% CIs in independent test set, prespecified subgroup analyses
- Development and validation of a computer vision model
- Patient-level partition into training (70%), validation (15%) and test (15%) sets
- Patient-level bootstrap 95% CIs in independent test set
- Prespecified subgroup analyses
Performance
As reported in the abstract
Detection of hand synovitis, independent test set
| Test or model | AUROC | Accuracy | Sensitivity | Specificity |
|---|---|---|---|---|
| ConvNeXt | 0.852 (95% CI: 0.802, 0.896) | 0.79 (95% CI: 0.75, 0.83) | 0.76 (95% CI: 0.67, 0.85) | 0.80 (95% CI: 0.75, 0.84) |
ConvNeXt
AUROC0.852 (95% CI: 0.802, 0.896)
Accuracy0.79 (95% CI: 0.75, 0.83)
Sensitivity0.76 (95% CI: 0.67, 0.85)
Specificity0.80 (95% CI: 0.75, 0.84)
Further findings
| Outcome | EfficientNetV2 | Ensemble |
|---|---|---|
| Validation AUROC | ||
| AUROC | 0.831 | 0.864 |
- Validation AUROC: ConvNeXt V2 AUROC 0.856
- Validation AUROC: ensemble α = 0.24
Validation AUROC
- Ensemble: modest incremental gain over ConvNeXt
Prespecified subgroups (deformity status, age, sex)
- Model performance remained stable across all prespecified subgroups, including patients with hand deformities
Original publication
Detecting inflammatory arthritis in hand smartphone photographs: development and validation of a computer vision model in clinical settings. Rheumatology (Oxford). Published 2 June 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →