Table 4.
Results of five regression models predicting FVC using best features and AAdiff.
| Regression Model | R2 | RMSE |
|---|---|---|
| Linear | 75.26 ± 7.14 | 0.4456 ± 0.0693 |
| Lasso (α = 0.0001) |
75.27 ± 7.12 | 0.4456 ± 0.0692 |
| Ridge (α = 0.4) |
75.28 1 7.12 | 0.4455 ± 0.0691 |
| Elastic Net (α = 0.0025) |
75.38 ± 6.98 | 0.4448 ± 0.0690 |
| Bayesian Ridge | 75.28 ± 7.13 | 0.4455 ± 0.0692 |
The models were developed using the four best features (height, sex, and weight at age 18 and FVC at age 10) with AAdiff as predictors of FVC. Here, AAdiff = AA at 18 ‒ AA at 10, R2 = average goodness-of-fit measure for regression models represented as a percentage and RMSE = average root mean squared error.