Variables | Accuracy | Precision | Recall | F1-score |
---|
Sociodemographic | 61% ± 7% | 56% ± 9% | 95% ± 6% | 70% ± 7% |
Network statistics data | 65% ± 5% | 66% ± 6% | 90% ± 6% | 76% ± 2% |
Combined | 74% ± 3% | 84% ± 7% | 79% ± 8% | 81% ± 2% |
- Table 4 shows the results of the machine learning model using sociodemographic/clinical variables and network statistics. Sociodemographic variables included age, sex, Black race, marital status, education, employment status, self-reported health, dialysis vintage, whether they would accept a living donation, and whether they would accept a deceased donation. The network variables included degree centrality, eigenvector centrality, closeness centrality, betweenness centrality, and clustering. The model measure are reported and there standard deviations are reported as percentages