Table 1 Misclassification rates of the Framingham Cohort 1 models predicting hypertension at visit 1.
From: Screening the genome to detect an association with hypertension
ModelA | Number of End-nodes | Cohort 1 Observed Error | Cohort 1 Cross-validation | Cohort 1 Model Applied to Cohort 2 |
|---|---|---|---|---|
1) No Splits | 1 | 22.1% | 22.1% | 20.2% |
2) All M & E | 3 | 18.5% | 23.1% | 21.6% |
3) All M | 5 | 17.4% | 26.9% | 22.7% |
4) All E, subset of M | 3 | 18.5% | 23.1% | 21.6% |