| CPC G16H 50/30 (2018.01) [A61B 8/02 (2013.01); A61B 8/0866 (2013.01); A61B 8/5223 (2013.01); G06N 20/00 (2019.01)] | 20 Claims |
|
1. A system, comprising:
a processor that executes computer-executable components stored in a computer-readable memory, the computer-executable components comprising:
a receiver component that accesses training data comprising a first plurality of medical feature collections associated with previously pregnant patients, wherein the medical feature collections are selected from a group of distinct medical feature collections comprising a biometric data medical feature collection, a demographic data medical feature collection, a fetal characteristics medical feature collection, and a medical facility characteristics medical feature collection;
a training component that:
for each of the first plurality of medical feature collections, trains, using back propagation, a first machine learning model corresponding to the medical feature collection to generate an embedded feature that represents a dimensionally reduced version of the medical feature collection; and
trains, using the respective embedded features generated by the first machine learning models and back propagation, a second machine learning model to generate risk scores indicative amounts of risk to health of at least one of pregnant patients or fetuses of the pregnant patients in at least one of performing caesarian-sections or waiting to give birth naturally;
wherein the receiving component accesses a second plurality of medical feature collections associated with a currently pregnant patient;
an embedding component that generates, via execution of the first trained machine learning models using the second plurality of medical feature collections, a plurality of embedded features associated with the currently pregnant patient;
a risk component that generates, via execution of the second trained machine learning model using the plurality of embedded features associated with the currently pregnant patient, at least one risk score associated with at least one of the currently pregnant patient or a fetus of the currently pregnant patient; and
an execution component that compares the at least one risk score to at least one predetermined threshold and that recommends, based on the comparison, performing the caesarian-section on the currently pregnant patient or waiting for the currently pregnant patient to give birth naturally.
|