A diagnostic prediction model for anti-neutrophil cytoplasmic antibody associated vasculitis combined with glomerulonephritis based on machine learning algorithm
ObjectiveTo construct a diagnostic model for anti-neutrophil cytoplasmic antibody （ANCA）-associated vasculitis （AAV） combined with glomerulonephritis using both random forest （RF） and artificial neural network algorithms.MethodsDatasets were downloaded from the GEO and Array Express databases （GSE10...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
Editorial Department of Journal of Clinical Nephrology
2025-02-01
|
| Series: | Linchuang shenzangbing zazhi |
| Subjects: | |
| Online Access: | http://www.lcszb.com/article/doi/10.3969/j.issn.1671-2390.2025.02.001 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | ObjectiveTo construct a diagnostic model for anti-neutrophil cytoplasmic antibody （ANCA）-associated vasculitis （AAV） combined with glomerulonephritis using both random forest （RF） and artificial neural network algorithms.MethodsDatasets were downloaded from the GEO and Array Express databases （GSE108113 and GSE104948 as training sets， E-MTAB-1944 as validation sets）. Differentially expressed genes （DEGs） between AAV combined with glomerulonephritis samples and normal controls were identified， followed by GO and KEGG enrichment analyses and protein-protein interaction （PPI） network analysis. RF and artificial neural network algorithms were jointly used to further screen characteristic genes. Diagnostic models were constructed and validated.ResultsA total of 380 DEGs were identified， involving 194 significantly up-regulated and 186 down-regulated ones. The enrichment analyses showed that DEGs were significantly enriched in the immune response and metabolic processes. EHHADH， CCL2， FN1， IL1B， VAV1， CXCR4， CCL5， and CD44were core genes in the PPI network. The RF algorithm screened out 15 characteristic genes， and the artificial neural network algorithm calculated the weight of each characteristic gene and successfully constructed a diagnostic model. The model had significant predictive power， with an area under the curve （AUC） of 1.000. The AUC of the validation cohort was 0.808， further confirming the accuracy of the model.ConclusionsCharacteristic biomarkers of AAV combined with glomerulonephritis are identified by machine learning algorithms， which are used to build a diagnostic model. The model can serve as a reliable reference for early diagnosis of the disease and provide a new perspective for the study of pathogenesis. |
|---|---|
| ISSN: | 1671-2390 |