mpm.rbm: Morphonode default Robust Binomial Model (RBM) endemble

In Morphonodepredictivemodel/morphonode: The Morphonode Predictive Model for ultrasound signatures detection and malignancy prediction in vulvar cancer

Morphonode default Robust Binomial Model (RBM) endemble

Description

Logistic model generated by fitting a simulated dataset of 948 ultrasound profiles (440 malignant and 508 non-malignant), with robust bootstrap standard errors estimation (5000 bootstrap iterations), through the internal morphonode function boot.se. The input ultrasound feature dataset was dichotomized before model fitting (see dichotomize) to avoid parameter estimation biases due to very low frequency of the levels of some categorical ultrasound features. The RBM is used to estimate the malignancy risk, (R) providing a continuous measure of malignancy (in contrast to the dichotomous prediction of the RFC ensemble). Considering the expected simulated phenotype (y) as the ground truth, two optimal malignancy risk cutoffs were estimated, defining three risk levels: low (R < 0.23), moderate (0.23 <= R <= 0.29), and high (R > 0.29).

Usage

mpm.rbm

Format

"mpm.rbm" is a list of 3 objects:

1. "coef", a data.frame reporting bootstrap-based estimations: ultrasound feature (Variable), log(odds ratio) (Estimate), bootstrap standard errors (se.boot), confidence interval lower bound (lower), confidence interval upper bound (upper), confidence level (conf.level), bootstrap estimation method (method), z-score (z), 2-sided p-value (P);

2. "model", R formula representing the fitted model;

3. "fit", MLE-based fitted model object of class glm.

References

Fragomeni SM, Moro F, Palluzzi F, Mascilini F, Rufini V, Collarino A, Inzani F, Giacò L, Scambia G, Testa AC, Garganese G (2022). Evaluating the risk of inguinal lymph node metastases before surgery using the Morphonode Predictive Model: a prospective diagnostic study. Ultrasound xx Xxxxxxxxxx xxx Xxxxxxxxxx. 00(0):000-000. <https://doi.org/00.0000/00000000000000000000>

Examples

# Create a simulated malignant ultrasound profile
x <- new.profile(us.simulate(y = 1))

# Lauch the Morhonode Predictive Model
u <- us.predict(x)

Morphonodepredictivemodel/morphonode documentation built on Feb. 15, 2023, 4:51 a.m.