mpm.us: Morphonode simulated ultrasound feature dataset

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

Morphonode simulated ultrasound feature dataset

Description

A simulated dataset of 948 ultrasound profiles (440 malignant and 508 non-malignant) used to build the default ensemble random forest classifier (RFC) and the robust binomial model (RBM). This dataset was generated as a 4-fold expansion of the original ultrasound feature dataset of 237 groin samples (75 malignant and 162 non-malignant) from Fragomeni et al. (2022), using the morphonode function us.simulate.

Usage

mpm.us

Format

"mpm.us" is a data.frame of 948 rows (simulated ultrasound profiles) and 18 columns, including: a progressive number used as unique profile identifier (ID), 14 ultrasound features used for RFC and RBM building, expected simulation phenotype used as ground truth (y = 0: non-malignant, 1: malignant), metastatic risk signature associated to each simulated ultrasound profile (signature), subject-level prediction error calculated as Brier score (E).

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

# Default simulated dataset phenotype frequencies
print(table(mpm.us$y))

# Default simulated dataset signatures frequencies
print(table(mpm.us$signature))

# Default simulated dataset prediction error quartiles
print(quantile(mpm.us$E))

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