US 12,133,741 B2
Predicting time to vaginal delivery
Anish J. Shah, Fort Lee, NJ (US)
Assigned to Birth Model, Inc., Fort Lee, NJ (US)
Filed by Birth Model, Inc., Fort Lee, NJ (US)
Filed on Apr. 20, 2023, as Appl. No. 18/137,099.
Application 18/137,099 is a continuation of application No. 17/887,875, filed on Aug. 15, 2022, granted, now 11,664,100.
Claims priority of provisional application 63/233,859, filed on Aug. 17, 2021.
Prior Publication US 2023/0260609 A1, Aug. 17, 2023
Int. Cl. A61B 5/00 (2006.01); G06N 20/00 (2019.01); G16H 10/60 (2018.01); G16H 50/20 (2018.01); G16H 50/70 (2018.01)
CPC A61B 5/4343 (2013.01) [G06N 20/00 (2019.01); G16H 10/60 (2018.01); G16H 50/20 (2018.01); G16H 50/70 (2018.01)] 20 Claims
OG exemplary drawing
 
1. A method performed by one or more computers, the method comprising:
receiving a request to predict a time to vaginal delivery of an infant by a patient; and
in response to receiving the request, generating a sequence of multiple delivery time predictions for the time to vaginal delivery of the infant by the patient, comprising, for each of a plurality of time points in a sequence of multiple time points:
determining, based at least in part on automated and continuous monitoring of real-time sensor data that is generated by one or more sensors and that characterizes a current physiological state of the patient, that a criterion for generating a new delivery time prediction for the time to vaginal delivery of the infant using a delivery time machine learning model has been satisfied; and
in response to determining, based at least in part on automated and continuous monitoring of real-time sensor data that is generated by one or more sensors and that characterizes the current physiological state of the patient, that the criterion for generating the new delivery time prediction for the time to vaginal delivery of the infant using the delivery time machine learning model has been satisfied:
generating a current model input that characterizes a current state of the patient by automatically querying a database storing one or more electronic medical records of the patient;
processing the current model input using the delivery time machine learning model, in accordance with trained values of a set of model parameters of the delivery time machine learning model, to generate a prediction for the time to vaginal delivery of the infant, and
generating a notification that indicates the prediction for the time to vaginal delivery of the infant;
wherein training the delivery time machine learning model comprises:
obtaining a set of training samples, wherein each training sample comprises: (i) a model input to the delivery time machine learning model that characterizes features of an individual, and (ii) a target delivery time that defines a time to vaginal delivery of the individual; and
training the delivery time machine learning model on the set of training samples by a machine learning technique, the training comprising, for each training sample:
training the delivery time machine learning model to minimize an error between: (a) a predicted delivery time that is generated by processing the model input of the training sample using the delivery time machine learning model, and (b) the target delivery time for the training sample;
wherein training the delivery time machine learning model further comprises:
dynamically monitoring one or more electronic medical record databases to determine when new electronic medical record data has been added to the one or more electronic medical records databases that enables the generation of new training samples for training the delivery time machine learning model;
automatically generating new training samples for training the delivery time machine learning model as new electronic medical record data is added to the one or more electronic medical record databases; and
re-training the delivery time machine learning model in response to determining that at least a threshold number of new training examples have been generated since the delivery time machine learning model was last trained.