Predictive machine learning model for preeclampsia using artificial neural networks

Abstract

Approaches are disclosed that may include generating and / or using a predictive machine learning classifier that includes one or more artificial neural networks. The classifier is configured to output a prediction regarding developing preeclampsia (e.g., early-onset preeclampsia) during the patient's current pregnancy based on health characteristics and one or more DNA metrics. The health characteristics and DNA metrics, such as total cell-free DNA (cfDNA) and fetal fraction (FF), may be obtained during routine non-invasive or minimally invasive prenatal screening. The predictive machine learning classifier may be generated by applying deep learning techniques to data regarding subjects in a cohort. The data may include characteristics corresponding to previous pregnancy outcomes, health indicators, one or more cfDNA measurements. Early pregnancy risk assessment for preeclampsia may identify patients most likely to benefit from a preventive treatment protocol with minimal or low levels of intervention.

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This patent application claims priority to U.S. Provisional Patent Application No. 63 / 324,469, filed March 28, 2022, and U.S. Provisional Patent Application No. 63 / 400,380, filed August 23, 2022, each of which is incorporated by reference in its entirety herein.

[0002] The present technology relates generally to using artificial intelligence to predict health outcomes in subjects during pregnancy, and more specifically to machine learning models for identifying patients at risk for developing preeclampsia during a current pregnancy, e.g., using patient characteristics collected non-invasively or minimally invasively during routine visits to a healthcare provider. [Background technology]

[0003] Preeclampsia is a major contributor to maternal mortality and morbidity. Early pregnancy risk assessment for early preeclampsia (<37 weeks gestation) can identify patients who are likely to benefit from low-dose aspirin. Low-dose aspirin initiated between 12-16 weeks gestation has been shown to reduce the risk of early-onset preeclampsia in high-risk patients. Accurate early pregnancy assessment of preeclampsia risk allows for identification of patients most likely to benefit from initiation of aspirin between 12-16 weeks gestation, where there is evidence of efficacy. Risk assessment is also important to guide appropriate pregnancy care pathways and surveillance.

[0004] Current United States Preventive Services Task Force (USPSTF) guidelines identify our limited clinical ability to identify patients at risk for preeclampsia and identify more accurate early pregnancy risk assessment as a priority. Previous approaches to predict preeclampsia have used patient demographic and clinical characteristics, ultrasound measurements, and biomarkers. However, the more expensive, difficult to obtain, invasive, and / or non-routine the health data required to predict early preeclampsia are, the more likely it is that at least some high-risk patients will not be identified because such health data are less commonly available. Summary of the Invention

[0005] Various possible embodiments of the disclosed approach relate to a method of applying a predictive machine learning classifier by a computing system to generate a prediction of a patient who will develop a preeclampsia condition, such as early or late preeclampsia, or other pregnancy outcome during a current pregnancy. The predictive machine learning classifier may include one or more artificial neural networks. Generating the prediction may include obtaining patient data by at least one of: (A) performing a non-invasive prenatal screening (NIPS) on the patient, and / or (B) receiving data from a health record system by the computing system via a communication network. The patient data may include (i) a set of health characteristics, and / or (ii) one or more cell-free DNA (cfDNA) metrics. As used herein, a cfDNA metric is any measurement, derived value, or calculated value based on data generated from one or more cfDNA analyses. Examples of cfDNA metrics include, but are not limited to, total cfDNA, fetal fraction (FF), SNP data from cfDNA screening, and the like. Generating a prediction may include feeding patient data to a predictive machine learning classifier to obtain a prediction of the patient developing early preeclampsia during a current pregnancy. The method may include providing a prediction for patient care by a computing system. Providing a prediction may include at least one of transmitting the prediction by the computing system via a communication network to at least one of a health record system and / or a computing device associated with a health care provider. The predictive machine learning classifier may be trained by applying one or more machine learning techniques to data regarding subjects in a cohort of pregnant subjects. In various possible implementations, the one or more machine learning techniques may include, for example, deep learning, logistic regression, random forest, and / or gradient boosting.The data may include, for each pregnant subject, a plurality of: (A) a first feature set indicative of one or more outcomes of one or more previous pregnancies of the pregnant subject; (B) a second feature set including one or more health indicators for the pregnant subject; and / or (C) a third feature set based on one or more cfDNA measurements for the pregnant subject. As used herein, health indicators (used interchangeably with health characteristics or health data) include values ​​(measured or derived) that correlate with health quality or wellness level, including but not limited to blood pressure, body mass index (BMI), weight, biomarkers (e.g., blood levels of various compounds), and the like. As used herein, patient characteristics (used interchangeably with patient data) may include or be health indicators and / or other information about the patient that does not necessarily indicate health quality or health level by itself. As a non-limiting example, sex at birth may be a patient characteristic.

[0006] In various embodiments, all of the patient data used to obtain the prediction may be collected non-invasively during one or more routine medical visits of the patient. In various embodiments, the patient data used to obtain the prediction may exclude uterine artery Doppler and / or other ultrasound data. In various embodiments, the preeclampsia is early preeclampsia or early onset preeclampsia. In various embodiments, the method may include generating, by a computing system, a training data set based on the first, second, and / or third feature sets. In various embodiments, the method may include training, by a computing system, a predictive machine learning classifier using data for subjects in a cohort of pregnant subjects. In various embodiments, the predictive machine learning classifier may include, for example, any combination of one or more rectified linear units (reLu), sigmoid linear units (SiLU), softmax functions, and the like. In various embodiments, the predictive machine learning classifier may include a hidden layer having the same dimensionality as the input layer. In various embodiments, the predictive machine learning classifier may include multiple hidden layers and may have varying dimensionality. In various embodiments, the one or more deep learning techniques may include using one or more regularizers (e.g., an L2 regularizer).

[0007] Various possible embodiments of the disclosed approach relate to a method that includes generating a predictive machine learning classifier for preeclampsia in a pregnant patient. The predictive machine learning classifier may include one or more artificial neural networks. Generating the predictive machine learning classifier may include generating, by a computing system, a training dataset using data on subjects in a cohort of pregnant subjects. The training dataset may include, for each pregnant subject, more than one of: (A) a first feature set indicative of one or more outcomes of one or more previous pregnancies of the subject; (B) a second feature set corresponding to one or more health indicators of the subject; and / or (C) a third feature set corresponding to one or more cfDNA metrics. The method may include applying, by the computing system, one or more deep learning techniques to the training dataset to generate a predictive machine learning classifier. The predictive machine learning classifier may be configured to receive as inputs the patient's health characteristics and one or more cfDNA metrics and output a prediction regarding the development of preeclampsia. The patient's health characteristics or other patient data may be obtained, for example, from or via an electronic medical record system. The method may include providing, by a computing system, a predictive machine learning classifier to generate a patient-specific index for the individual patient. Providing the predictive machine learning classifier may include at least one of: (i) transmitting, by the computing system over a communications network, the predictive machine learning classifier to a second computing system, and / or (ii) storing the predictive machine learning classifier in a non-transitory computer-readable storage medium accessible for subsequent use by a healthcare provider. The second computing system may be, for example, an electronic medical record system.

[0008] In various exemplary implementations, the predictive machine learning classifier may include one or more hidden regularized linear units (reLu). In various implementations, the predictive machine learning classifier may include a hidden layer having the same dimension as the input layer. In various implementations, applying the one or more deep learning techniques may include using one or more regularizers. In various implementations, the one or more regularizers may include an L2 regularizer. In various implementations, the method may include using the predictive machine learning classifier to evaluate the patient during the current pregnancy by performing a non-invasive prenatal screening (NIPS) on the patient, by feeding data to the predictive machine learning classifier to obtain a prediction of the development of preeclampsia during the current pregnancy (the data includes data from the NIPS), and / or by using the prediction in the delivery of medical care to the patient. In various implementations, all of the data provided to the predictive machine learning classifier to obtain the prediction may be collected non-invasively or minimally invasively (e.g., NIPS or blood drawing for biomarkers) during one or more routine medical visits of the patient. In various embodiments, the method may further include using a predictive machine learning classifier to evaluate the patient during the current pregnancy by receiving data corresponding to one or more previous pregnancies, one or more health characteristics, and one or more cfDNA metrics of the patient from a health record system, feeding the data to the predictive machine learning classifier to obtain a prediction of the occurrence of pre-term pre-eclampsia during the current pregnancy, and / or using the prediction in the delivery of medical care to the patient. In various embodiments, all data input to the predictive machine learning classifier to obtain a prediction may be collected non-invasively or minimally invasively during one or more regular medical visits of the patient. In various embodiments, the pre-term pre-eclampsia may be early-onset pre-term pre-eclampsia.

[0009] Various possible embodiments of the disclosed approach relate to a computing system configured to apply a predictive machine learning classifier to generate a prediction of a patient who will develop an early preeclampsia condition during a current pregnancy. The predictive machine learning classifier may include one or more artificial neural networks. The computing system may include one or more processors configured to receive patient data including a set of health characteristics and one or more cfDNA measurements over a communication network during the patient's current pregnancy. The one or more processors may be configured to feed the patient data to the predictive machine learning classifier to obtain a prediction of the patient who will develop early preeclampsia during a current pregnancy. Data used for and / or fed to the predictive machine learning classifier may be obtained from or via one or more electronic health record systems. The one or more processors may be configured to provide a prediction for the care of the patient. Providing the prediction may include transmitting the prediction over the communication network to at least one of a health record system (e.g., an electronic medical record) and / or a computing device associated with a health care provider. The predictive machine learning classifier may have been trained by applying one or more deep learning techniques to data regarding subjects in a cohort of pregnant subjects. The training data may include, for each pregnant subject, a plurality of (A) a first feature set indicative of one or more outcomes of the pregnant subject's one or more previous pregnancies, (B) a second feature set corresponding to one or more health indicators of the pregnant subject, and / or (C) a third feature set corresponding to one or more cell-free DNA metrics for the pregnant subject.

[0010] In various embodiments, all of the patient data used to obtain the prediction may have been collected non-invasively during one or more routine medical visits of the patient. In various embodiments, the patient data used to obtain the prediction may exclude ultrasound data. In various embodiments, the pre-eclampsia is early-onset pre-eclampsia. In various embodiments, the one or more processors are configured to generate a training data set and apply deep learning techniques to the training data set to generate a predictive machine learning classifier. In various embodiments, the one or more processors are configured to receive at least a subset of the health data from a health record system via a communication network. In various embodiments, the one or more processors are configured to receive at least a subset of the health data via a software application executing on a computing device associated with the health care provider.

[0011] Various embodiments of the present disclosure may relate to processes performed using the devices and / or systems disclosed herein.

[0012] Various embodiments of the present disclosure may relate to computing systems and / or devices for performing the processes disclosed herein. [Brief description of the drawings]

[0013] [Figure 1] 1 illustrates an exemplary system for implementing the disclosed techniques, according to various possible embodiments. [Diagram 2] 1 illustrates an exemplary modeling and prediction process, according to various possible embodiments. [Diagram 3] FIG. 1 shows a simplified block diagram of representative server and client computer systems that can be used to implement certain embodiments of the present disclosure. [Figure 4] 1 illustrates data used to generate and validate an exemplary model, according to a possible embodiment of the disclosed approach. [Diagram 5] 1 illustrates an exemplary production workflow according to a possible embodiment of the disclosed approach. [Figure 6] Figure 1 shows the significance of features and their relative contribution to the final prediction of early preeclampsia according to possible embodiments of the disclosed approach. The top (A) corresponds to a linear neural network (LNN) model, and the bottom (B) corresponds to a nonlinear neural network (NLNN) model. The average contribution and 95% confidence intervals are shown. The size and orientation of the bars represent the contribution of each single feature to the definition of patients positive for early preeclampsia. [Figure 7] Figure 1 shows the relative contribution of features to the significance and final prediction of early preeclampsia for reduced models with cfDNA features omitted, according to possible embodiments of the disclosed approach. The top (A) corresponds to the reduced LNN model, and the bottom (B) corresponds to the reduced NLNN model. The average contribution and 95% confidence interval are shown. The size and direction of the bars represent the contribution of each single feature to the definition of patients positive for early preeclampsia. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] The above and other features of the present disclosure will become apparent from the following description and appended claims, taken in conjunction with the accompanying drawings, in which: The present disclosure will be described with additional specificity and detail through the use of the accompanying drawings, with the understanding that these drawings illustrate only some embodiments according to the present disclosure and therefore should not be considered limiting of its scope.

[0015] It should be understood that certain aspects, modes, embodiments, variations, and features of the method are described below at various levels of detail to provide a substantial understanding of the present technology. It should be understood that the present disclosure is not limited to particular applications, methods, devices, or systems, each of which may, of course, vary. It should also be understood that the terminology used herein is for the purpose of describing particular embodiments only, and is not intended to be limiting.

[0016] Less interventional preventative treatments (such as low-dose aspirin) tend to be associated with a lower risk of complications or adverse outcomes, as well as lower costs, and are therefore generally preferred over more interventional treatments. Patient characteristics known to be associated with increased risk of preeclampsia are routinely collected as part of prenatal care. In addition, maternal and fetal placental cell-free DNA (cfDNA) results from early pregnancy that may be associated with increased risk of developing preeclampsia are frequently available as part of noninvasive prenatal screening (NIPS), the predominant screening test for fetal chromosomal abnormalities in the United States.

[0017] In a possible embodiment, an artificial neural network model is generated and / or used to predict early preeclampsia (preeclampsia before 37 weeks of pregnancy) and early onset preeclampsia (preeclampsia before 34 weeks of pregnancy) using routinely available and / or non-invasively obtained patient characteristics, such as health data available at the first prenatal visit and data from cfDNA screening.

[0018] 1, in a possible implementation, the system 100 may include a computing system 110 (which may be or may include one or more computing devices co-located with each other or remote from each other), a condition detection system 160, a clinical information system (CIS) 170 (used interchangeably with electronic medical record (EMR) system or electronic health record (EHR) system), a platform 175, and a treatment system 180. The computing system 110 (e.g., one or more computing devices) may be used to control and / or exchange signals and / or data with the condition detection system 160, the CIS 170, the platform 175, and / or the treatment system 180, either directly or via another component of the system 100. In a particular embodiment, the computing system 110 may be used to control and / or exchange data or other signals with the condition detection system 160, the CIS 170, the platform 175, and / or the treatment system 180. The computing system 110 may include one or more processors, and one or more volatile and non-volatile memories for storing captured, acquired, recorded, and / or generated computing code and data.

[0019] The computing system 110 may include a controller 112 configured to exchange control signals with the condition detection system 160, the CIS 170, the platform 175, the treatment system 180, and / or any components thereof, allowing the computing system 110 to be used, for example, to control the acquisition of patient data such as test results, the capture of images, the acquisition of signals by sensors, the positioning or repositioning of subjects and patients, the recording or acquisition of other subject or patient information, and the application of treatments.

[0020] The transceiver 114 enables the computing system 110 to exchange readings, control commands, and / or other data, wirelessly or wired, directly or via a network protocol, with the condition detection system 160, the CIS 170, the platform 175, and / or the treatment system 180, or components thereof. One or more user interfaces 116 enable the computing device 110 to receive user input (e.g., via a keyboard, touch screen, microphone, camera, etc.) and provide output to a user (e.g., via a display screen, audio speakers, etc.). The computing device 110 may further include one or more databases 118 for storing, for example, data obtained from one or more systems or devices, signals obtained via one or more sensors, biomarker signatures, etc. In some embodiments, database 118 (or a portion thereof) may alternatively or additionally be part of another computing device, either co-located or remotely located (e.g., via “cloud computing”), in communication with computing device 110, condition detection system 160, CIS 170, platform 175, and / or treatment system 180 or components thereof.

[0021] The condition detection system 160 may include a test system 162, which may be or include, for example, any system or device involved in analyzing samples (e.g., blood, plasma, or urine), recording patient data, and / or obtaining laboratory or other test results (e.g., DNA data). The imaging system 164 may be any system or device used to capture imaging data, such as a magnetic resonance imaging (MRI) scanner, a positron emission tomography (PET) scanner, a single photon emission computed tomography (SPECT) scanner, a computed tomography (CT) scanner, a fluoroscopy scanner, and / or other imaging devices and / or sensors. The sensors 166 may detect, for example, the patient's position or movement, physiological measurements such as organs, tissues, lung volumes, or cardiac activity / signals, or other conditions and / or conditions of the patient.

[0022] The treatment system 180 may include a treatment unit 180, which may be or may include, for example, a radiation source for external beam therapy (e.g., an orthovoltage x-ray machine, a cobalt 60 machine, a linear accelerator, a proton beam machine, a neutron beam machine, etc.) and / or one or more other treatment devices. The sensor 184 may be used by the treatment system 180 to evaluate and guide treatment (e.g., by detecting a level of emitted radiation, a patient condition or status, or other status or condition). In various implementations, the components of the system 100 may be rearranged or integrated into other configurations. For example, the computing system 110 (or components thereof) may be integrated with the condition detection system 160, the treatment system 180, and / or one or more of its components. The condition detection system 160, the treatment system 180, and / or components thereof may be directed to a platform 175 on which a patient, subject, or sample may be located (to test the sample, image the subject, apply a treatment or therapy to the subject, detect activity and / or movement of the subject in a stress test, etc.). In various embodiments, the platform 175 may be movable (e.g., using any combination of motors, magnets, etc.) to allow positioning and repositioning of the sample and / or subject (e.g., fine adjustments to compensate for subject or patient movement, or to position the patient for scanning of different regions of interest). The platform 175 may include its own sensors to detect a condition or status of the sample, patient, and / or subject.

[0023] Computing system 110 may include, for example, a tester and imager 120 configured to direct laboratory testing, image capture, and acquisition of test and / or imaging data. Tester and imager 120 may include an image generating device that may convert or transform raw imaging data from condition detection system 160 into usable medical images or another form for analysis. Computing system 110 may include a test and image analyzer 122 configured, for example, to analyze raw test results and generate associated metrics, identify features within the image or imaging data, or otherwise utilize the test or test data and / or image or imaging data.

[0024] The data acquisition unit 124 may retrieve, acquire, or otherwise obtain various data used, for example, to train a model using subject data or to apply a model trained using patient data. The data acquisition unit 124 may obtain test data stored in the CIS 170, for example. The interaction unit 126 may interact (e.g., via the user interface 116) with a user (e.g., a patient and / or a healthcare provider) to obtain information needed for the model or to provide information to the user. In certain embodiments, the data acquisition unit 124 may obtain data from a user via the interaction unit 126.

[0025] The machine learning model trainer 130 ("ML model trainer") may be configured to train the machine learning models used herein, as further described below. The ML model trainer may include, for example, a training dataset generator that processes various data to generate one or more training datasets used to train the models described herein. In certain embodiments, as more data and outcomes become available, the model retrainer 134 may further train the pre-trained model to improve or otherwise update or modify the model. The ML model trainer may apply various machine learning techniques to the training dataset. The machine learning modeler 140 ("ML modeler") may be configured to apply the trained machine learning model to specific patient data. A feature selector 142 may be configured to select which features are fed to the model to obtain a prediction (e.g., the likelihood of developing late-term preeclampsia, early-term preeclampsia, and / or early-onset preeclampsia). The feature selector 142 may select features based on, for example, which features were previously used to train the ML model(s) being used, the parameters of the model(s), and / or how influential the feature was in predicting the outcome (e.g., feature coefficients). The feature selector 142 may additionally or alternatively select features based on what data (e.g., test results) is available for the patient. For example, if data corresponding to one feature in the "Tier 1" feature set (e.g., the "top 5" most influential features with the 5 highest coefficients) is not available for the patient, two features in the "Tier 2" feature set (e.g., the features with the 6th-10th highest coefficients) may be selected instead. The feature extractor 144 may obtain values ​​for the selected features from various data sources. An output and reporting unit 150 may transmit the model or its predictions (e.g., to a medical professional, a patient, etc.) or generate and provide a report based on the model or its predictions.The report may include, for example, the model output (e.g., a prediction) along with information regarding the basis for the prediction, and guidelines and / or suggested recommendations and next steps based on the prediction to identify factors that most influence a low or high likelihood or other outcome for the clinician and to guide the clinician and the patient.

[0026] Referring to FIG. 2, an exemplary process 200 according to various possible embodiments is shown. Various elements of the process 200 may be implemented by or through the system 100 or its components. The process 200 may start with model training (left side of FIG. 2), which may be implemented by or through the computing system 110 if a model is not already available (e.g., in the database 118) or if an additional model is generated or updated by training or retraining with new training data (205). Alternatively, the process 200 may start with the use (application) of a model (right side of FIG. 2) for a patient if a trained model is already available. If a suitable trained model is available, prediction may be implemented by or through the computing system 110. In various embodiments, the process 200 may include both model training (e.g., steps 210-225) followed by prognosis prediction (e.g., steps 250-270).

[0027] At 210, health data may be obtained for the subjects. This may be obtained by or via, for example, condition detection system 160 and / or CIS 170, for a cohort of, for example, 10,000, 11,000, 12,000, 13,000, 14,000, 15,000, or more subjects. This may represent, for example, multiple years of treatment (e.g., multiple pregnancies). The health data needs to have enough data (e.g., across multiple pregnancies) to capture and describe the various features and be able to build a generalizable model. The laboratory test or imaging data may be transformed, converted, or otherwise analyzed by test and image analyzer 122. Step 215 includes extracting feature values ​​corresponding to the subjects in the cohort from (or based on) the health data.

[0028] At 220, one or more datasets (e.g., training datasets) may be generated using the extracted feature values ​​corresponding to the subject's health data. This may be performed, for example, by or through the ML model trainer 130, or more specifically, the training and test dataset generator 132. At 225, the one or more datasets may be used to train and test a model for making predictions. The model may be stored (e.g., in database 118) for subsequent use. Process 200 may end (290) or may proceed to step 250 for use in prognosis predictions. (As represented by the dotted line from step 225 to step 265, the model may then be used to generate and use predictions.)

[0029] At 250, patient health data may be acquired and analyzed (e.g., by or via condition detection system 160 and / or CIS 170). The test data may correspond to various laboratory tests performed on the patient's samples. At 255, a set of features may be selected (e.g., by or via feature selector 142), and at 260, feature values ​​may be extracted from the patient's health data (e.g., by or via feature extractor 144).

[0030] At 265, the feature values ​​extracted from the patient's health data may be input into a predictive model (e.g., a machine learning classifier) ​​to predict the patient's prognosis. At 270, the predicted prognosis and factors underlying the prognosis may be used in the care of the patient. For example, the predicted prognosis may be used to plan for potential outcomes and / or identify potential preventative care. Additionally or alternatively, the predicted prognosis may be used to evaluate the patient's health and changes or trends therein, such as whether the patient is getting worse or improving as indicated by a change in prognosis over time from the ML modeler as new tests are performed or new health data otherwise becomes available.

[0031] The process 200 may end (290) or may return to step 250 for subsequent planning (e.g., after performing another test or administering a procedure) based on changes in the patient's condition.

[0032] Various operations described herein can be implemented on computer systems having a variety of configurations. Figure 3 shows a simplified block diagram of a representative server system 300 (e.g., computing system 110) and client computer system 314 (e.g., computing system 110, condition detection system 160, CIS 170, and / or treatment system 180) that can be used to implement various embodiments of the present disclosure. In various embodiments, the server system 300 or a similar system can implement a service or server described herein, or a portion thereof. The client computer system 314 or a similar system can implement a client described herein.

[0033] The server system 300 may have a modular design incorporating a number of modules 302 (e.g., blades in a blade server embodiment), and although two modules 302 are shown, any number may be provided. Each module 302 may include a processing unit(s) 304 and local storage 306.

[0034] The processing unit(s) 304 may include a single processor (which may have one or more cores), or multiple processors. In some embodiments, the processing unit(s) 304 may include a general-purpose primary processor, as well as one or more special-purpose co-processors, such as graphic processors, digital signal processors, and the like. In some embodiments, some or all of the processing units 304 may be implemented using customized circuitry, such as application specific integrated circuits (ASICs) or field programmable gate arrays (FPGAs). In some embodiments, such integrated circuits execute instructions stored in the circuitry itself. In other embodiments, the processing unit(s) 304 may execute instructions stored in local storage 306. Any type of processor in any combination may be included in the processing unit(s) 304.

[0035] The local storage 306 may include volatile storage media (e.g., conventional DRAM, SRAM, SDRAM, etc.) and / or non-volatile storage media (e.g., magnetic or optical disks, flash memory, etc.). The storage media incorporated in the local storage 306 may be fixed, removable, or upgradeable as desired. The local storage 306 may be physically or logically divided into various sub-units such as a system memory, a read-only memory (ROM), and a persistent storage device. The system memory may be a read-and-write memory device, such as a dynamic random access memory, or a volatile read-and-write memory. The system memory may store some or all of the instructions and data required by the processing unit(s) 304 at run time. The ROM may store static data and instructions required by the processing unit(s) 304. The persistent storage device may be a non-volatile read-and-write memory device that may store instructions and data even when the module 302 is powered off. As used herein, the term "storage medium" includes any medium capable of storing data indefinitely (subject to constraints such as overwriting, electrical disturbance, and power loss), and does not include carrier waves and transitory electronic signals propagated over wireless or wired connections.

[0036] In some embodiments, the local storage 306 may store one or more software programs executed by the processing unit(s) 304, such as an operating system and / or programs implementing various server functions, or any of the systems or devices described herein.

[0037] "Software" generally refers to a sequence of instructions that, when executed by the processing unit(s) 304, causes the server system 300 (or a portion thereof) to perform various operations and thus defines one or more specific machine embodiments that execute and perform the operations of the software program. The instructions can be stored as firmware residing in a read-only memory and / or as program code stored in a non-volatile storage medium that can be loaded into a volatile working memory for execution by the processing unit(s) 304. The software can be implemented as a single program or a collection of separate programs or program modules that interact as desired. From the local storage 306 (or non-local storage, described below), the processing unit(s) 304 can retrieve program instructions to execute and data to process in order to perform the various operations described above.

[0038] In some server systems 300, multiple modules 302 may be interconnected via a bus or other interconnect 308 to form a local area network that supports communication between the modules 302 and other components of the server system 300. The interconnect 308 may be implemented using a variety of technologies, including server racks, hubs, routers, etc.

[0039] A wide area network (WAN) interface 310 may provide data communication capabilities between the local area network (interconnect 308) and a larger network, such as the Internet. Conventional or other active technologies may be used, including wired (e.g., Ethernet, IEEE 802.3 standard) and / or wireless technologies (e.g., Wi-Fi, IEEE 802.11 standard).

[0040] In some embodiments, the local storage 306 is intended to provide working memory for the processing unit(s) 304 and provide fast access to programs and / or data being processed while reducing traffic on the interconnect 308. Storage for larger amounts of data can be provided on a local area network by one or more mass storage subsystems 312, which can be connected to the interconnect 308. The mass storage subsystems 312 can be based on magnetic, optical, semiconductor, or other data storage media. Direct attached storage, storage area networks, network-attached storage, etc. can be used. Any data store or other collection of data described herein as generated, consumed, or maintained by a service or server can be stored in the mass storage subsystem 312. In some embodiments, additional data storage resources can be accessible via the WAN interface 310 (potentially increasing latency).

[0041] The server system 300 may operate in response to requests received via the WAN interface 310. For example, one of the modules 302 may implement a monitoring function and assign separate tasks to the other modules 302 in response to received requests. Conventional work allocation techniques may be used. Once a request is processed, a result may be returned to the requester via the WAN interface 310. Such operations may generally be automated. Furthermore, in some embodiments, the WAN interface 310 may connect multiple server systems 300 together to provide a scalable system capable of managing large volumes of activity. Conventional or other techniques for managing server systems and server farms (collections of cooperating server systems), including dynamic resource allocation and reallocation, may be used.

[0042] The server system 300 can interact with a variety of user-owned or user-operated devices over a wide area network such as the Internet. One example of a user-operated device is shown in FIG. 3 as a client computing system 314. The client computing system 314 can be implemented as a consumer device, such as, for example, a smartphone, other mobile phone, a tablet computer, a wearable computing device (e.g., smart watch, glasses), a desktop computer, a laptop computer, etc.

[0043] The client computing systems 314 can communicate over the WAN interface 310. The client computing systems 314 can include traditional computer components such as a processing unit(s) 316, a storage device 318, a network interface 320, a user input device 322, and a user output device 324. The client computing systems 314 can be computing devices implemented in a variety of form factors, such as desktop computers, laptop computers, tablet computers, smartphones, other mobile computing devices, wearable computing devices, and the like.

[0044] The processor 316 and storage device 318 can be similar to the processing unit(s) 304 and local storage 306 described above. An appropriate device can be selected based on the demands placed on the client computing system 314, for example, the client computing system 314 can be implemented as a "thin" client with limited processing capabilities or as a high-powered computing device. The client computing system 314 can be provided with program code executable by the processing unit(s) 316 to enable various interactions with the server system 300 of the message management service, such as accessing messages, performing actions on messages, and other interactions described above. Some client computing systems 314 can also interact with the messaging service independent of the message management service.

[0045] The network interface 320 may provide a connection to a wide area network (e.g., the Internet) to which the WAN interface 310 of the server system 300 is also connected. In various embodiments, the network interface 320 may include a wired interface (e.g., Ethernet) and / or a wireless interface implementing various RF data communication standards, such as Wi-Fi, Bluetooth, or cellular data network standards (e.g., 3G, 4G, 5G, LTE, etc.).

[0046] The user input device 322 may include any device (or devices) by which a user can provide signals to the client computing system 314, which can be interpreted by the client computing system 314 as indicating a particular user request or information. In various embodiments, the user input device 322 may include any or all of a keyboard, a touchpad, a touchscreen, a mouse or other pointing device, a scroll wheel, a click wheel, a dial, a button, a switch, a keypad, a microphone, and the like.

[0047] The user output device 324 may include any device by which the client computing system 314 can provide information to a user. For example, the user output device 324 may include a display that displays images generated by or delivered to the client computing system 314. The display may incorporate various image generating technologies, e.g., liquid crystal display (LCD), light emitting diode (LED) including organic light emitting diode (OLED), projection system, cathode ray tube (CRT), etc., along with supporting electronics (e.g., digital-to-analog or analog-to-digital converters, signal processors, etc.). Some embodiments may include devices such as a touch screen that function as both an input device and an output device. In some embodiments, other user output devices 324 may be provided in addition to or instead of a display. Examples include indicator lights, speakers, tactile "display" devices, printers, etc.

[0048] Some embodiments include electronic components such as a microprocessor, storage and memory that store computer program instructions on a computer-readable storage medium. Many of the features described herein can be implemented as a process specified as a set of program instructions coded on a computer-readable storage medium. When these program instructions are executed by one or more processing units, they cause the processing unit(s) to perform various operations indicated in the program instructions. Examples of program instructions or computer code include machine code, such as generated by a compiler, and files containing higher level code that is executed by a computer, electronic component, or microprocessor using an interpreter. Through appropriate programming, the processing unit(s) 304 and 316 can provide various functions for the server system 300 and the client computing system 314, including any of the functions described herein as being performed by a server or client, or other functions related to a message management service.

[0049] It will be understood that the server system 300 and the client computing system 314 are exemplary and that variations and modifications are possible. Computer systems used in connection with the embodiments of the present disclosure may have other functionality not specifically described herein. Additionally, while the server system 300 and the client computing system 314 are described with reference to certain blocks, it will be understood that these blocks are defined for convenience of description and are not intended to imply a particular physical arrangement of components. For example, different blocks may, but need not, be located in the same facility, in the same server rack, or on the same motherboard. Additionally, the blocks need not correspond to physically distinct components. The blocks may be configured to perform various operations, for example, by programming a processor or by providing appropriate control circuitry, and the various blocks may or may not be reconfigurable depending on how the initial configuration is obtained. The embodiments of the present disclosure may be realized in a variety of apparatuses, including electronic devices implemented using any combination of circuitry and software. EXAMPLES

[0050] Exemplary Methods ANALYSIS:We conducted a secondary analysis of the Single-nucleotide-polymorphism (SNP)-based Microdeletion and Aneuploidy Registry (SMART). SMART was designed to evaluate the performance of SNP-based NIPS in the general pregnant population undergoing screening as part of clinical care. Between April 2015 and January 2019, 21 centers in six countries participated in the study. The study was restricted to singleton pregnancies, gestational age ≥9 weeks, and maternal age ≥18 years. Data on patient characteristics, NIPS results, serum screening results, ultrasound, and pregnancy outcomes were collected by the local study coordinator.

[0051] Study Population: Referring to Figure 4, a total of 20,887 women participated in SMART. For this analysis, pregnancies were excluded if there was a fetal chromosomal or major structural abnormality, termination of pregnancy, lack of data on pregnancy outcome, participant withdrawal from the study (n=1604), or lack of data on preeclampsia status or gestational age at delivery (n=365). Participants were further excluded if data were missing for any of the variables required for model development. The final cohort used for model development included 17,520 participants, 671 with preeclampsia and 16,849 without preeclampsia. Of those with preeclampsia, 251 had early preeclampsia, 72 had early-onset preeclampsia, and 420 had late preeclampsia. Preeclampsia is a pregnancy-associated condition consisting of high blood pressure with or without evidence of organ dysfunction (e.g., liver and kidney failure), symptoms (e.g., headaches and visual disturbances), and, in severe cases, seizures ("eclampsia"). The presence of "preeclampsia" was based on the presence of a clinical diagnosis of preeclampsia recorded in the participant's medical record.

[0052] Patient characteristics: Patient characteristics examined in the analysis included maternal age, body mass index (BMI), maternal height, maternal weight, nationality, race / ethnicity, parity, use of in vitro fertilization, chronic hypertension, pregestational diabetes, and smoking during pregnancy. Features derived from SNP-based NIPS included total cfDNA and fetal fraction (FF). Patient characteristics were compared between patients with and without preeclampsia. Comparisons of continuous data were performed using the Wilcoxon rank sum test, and comparisons of independent categorical data were performed using the chi-square test. P values ​​were adjusted for multiple comparisons using the Holm method.

[0053] Outcome measures: The primary outcome was early preeclampsia, defined as preeclampsia at birth before 37 weeks' gestation. Secondary outcomes were early-onset preeclampsia (preeclampsia at birth before 34 weeks' gestation) and late-preeclampsia (preeclampsia at birth ≥37 weeks' gestation).

[0054] Features in the model: Variables entered into the prediction model included 13 patient characteristics available from the SMART study data and identified as risk factors for the development of preeclampsia based on existing medical literature (Table 1). Two illustrative features obtained from SNP-based cfDNA screening were also included: total cfDNA and FF. FF was estimated by a probability model using SNPs considered to be homozygous in the maternal allele. The probability model used a maximum likelihood approach based on observed minor allele frequencies to determine the fetal fraction value that maximizes the likelihood function. Two features obtained from SNP-based cfDNA screening were also included: total cfDNA (maternal and fetal) and FF. Six of the 15 features were coded as continuous variables, and the remaining 10 features were coded as binary (Table 1).

[0055] [Table 1]

[0056] Including race and ethnicity in predictive modeling has the potential to perpetuate racial disparities. Thus, race and ethnicity were not included in the model development of this exemplary embodiment. However, given recent studies supporting the importance of race in preeclampsia prediction, a sensitivity analysis was performed to evaluate the effect of including race / ethnicity on the predictive accuracy of the model. Ethnicity information was added as a binary flag to the feature set in Table 1. Four ethnic groups (black, Asian, Latino, white) were modeled separately using the neural network model development and cross-validation protocol described in this method section. For example, for the black group, a binary flag was added to the feature list in Table 1. This flag was set to 1 for samples listed as "black" ethnicity and 0 for all other samples. Each group was examined in this way in turn, and the area under the receiver operating characteristic (AUROC) performance metric was calculated for each iteration. In some implementations, the model may be trained and tested (validated) using a common protocol based on, for example, stratified, repeated k-fold cross validation (SRKFCV).

[0057] Machine learning models for classification: First, a linear neural network (LNN) model was considered as the "baseline classifier". The LNN model is a neural network architecture that directly connects input features to a single sigmoid output neuron. In addition, several feed-forward dense neural network models (NLNN models) were considered, which incrementally add one or more hidden layers of rectified linear unit (reLu) units to the LNN model. The performance of the classifiers was evaluated on several metrics: AUC score, sensitivity, positive predictive value and negative predictive value at different values ​​of the screening positive rate metric.

[0058] Training and Internal Validation: Referring to Figure 5, an exemplary training and validation process 500 includes a series of steps described below. All neural network models in the exemplary implementation described with respect to Figure 5 below were implemented and optimized in keras.tensorflow using binary cross entropy as the loss function. The parameters of the networks were optimized by stochastic gradient descent with a constant learning rate of 1e-2 and momentum of 0.

[0059] In block 505, a model parameter accumulator and a score accumulator are initialized. The model parameter accumulator and the score accumulator are data structures in which results are stored at each iteration of a loop described below (blocks 515-550). In block 510, a variable i is initialized to indicate an iteration (here, the first iteration) of a series of steps that may be looped through until a value (n) is reached.

[0060] At block 515, there was a stratified split, where the dataset was randomly split into a training dataset and a validation dataset. In an exemplary embodiment, the split results in proportions of 0.9 and 0.1 corresponding to training and validation, respectively (i.e., 90% for training and 10% for testing). Here, the random split was also stratified with respect to outcome.

[0061] In block 520, a training data set and a test data set were generated using a step that takes into account imbalanced data. Only 1.4% of the study population developed pre-term pre-eclampsia, and therefore, a machine learning model applied to such a data set is typically expected to over-predict the majority class (i.e., patients without pre-term pre-eclampsia) and be relatively inefficient in predicting "rare" events associated with the minority class (i.e., patients with pre-term pre-eclampsia). In light of this, after the training and validation split (515), oversampling was applied to the positive samples in the training set via replacement with uniform random sampling to obtain a training set with the same number of positive and negative samples.

[0062] At block 525, the exemplary process 500 includes the steps of neural network scaling and noise injection. A standard scaler was applied to the oversampled training dataset to appropriately scale the training and validation datasets. Gaussian noise with mean equal to 0 and standard deviation equal to 1e-1 was added to the standardized training dataset. The standard scaler and Gaussian noise were applied only to continuous input features (Table 1). Noise injection was used as a form of both regularization and data augmentation.

[0063] The model was fitted to the oversampled training dataset in block 530. The fitting procedure was then performed using stochastic gradient descent with a constant learning rate of 1e-2 for 100 epochs and a mini-batch size of 32.

[0064] Relevant performance metrics were calculated on the validation dataset at block 535. Specifically, the predictive performance of the two classifiers was evaluated by calculating the area under the receiver operating characteristic (AUC) curve, as well as the sensitivity, positive predictive value, and negative predictive value for a fixed screen positivity rate.

[0065] In block 540, the model parameters were stored in the model parameter accumulator, and the scores were stored in the score accumulator. To repeat the steps within the loop n times, at 545, it is determined whether i (initially equal to 1) is equal to n. If not (for example, if i < n), in block 550, i is incremented upward, and process 500 loops back to block 515. If so (for example, if i = n), process 500 ends at block 555.

[0066] Internal validation: In exemplary process 500, blocks 515 - 530 were repeated 200 times (n = 200) to minimize or otherwise reduce the bias specific to a single instance of the random split in block 515. This Monte Carlo cross - validation generated 200 estimates for each of the performance metrics of interest, corresponding to 200 different random training - validation splits. As a result, the collected metrics are de facto random variables and can be analyzed as such. For each metric, the "expected behavior" of the prediction model was evaluated considering the mean value and its 95% confidence interval.

[0067] Network regularization: In addition to the noise injection described above (block 525), a L2 penalty that varies between 0.0 and 1e - 2 in magnitude was used to introduce regularization to the exemplary model via kernel_regularizer.

[0068] Hyperparameter optimization: Further, initially fixing the number of epochs and batch size in consideration, a partial but grid - based hyperparameter search was performed incrementally. Second, the number and size of the hidden layers, and three regularization methods, namely the magnitude of the L2 penalty in kernel regularization in the training dataset, the percentage in layer dropout, and the standard deviation in Gaussian noise injection, were investigated.

[0069] Model performance: Model performance was assessed by the prediction area under the curve (AUC) score and sensitivity for fixed screening positivity rates of 10%, 15% and 20% for primary and secondary outcomes.

[0070] Characteristics of the study population: Referring again to Figure 4, of the 17,520 participants included in the final analysis, 671 (3.8%) developed preeclampsia. Of these, 251 (1.4%) developed early preeclampsia, 72 (0.4%) developed early-onset preeclampsia, and 420 (2.4%) developed late preeclampsia. The characteristics of the 17,520 participants (671 with preeclampsia and 16,849 without preeclampsia) included in the model development are shown in Table 2. The group that developed preeclampsia had a similar maternal age distribution to the group that did not develop preeclampsia (34.3 years vs. 34.7 years, p=0.022). The BMI of 30 kg / m 2 The proportion of participants with > 18.8% was significantly higher in the preeclampsia group (40.8% vs. 18.8%, p<0.001). Participants in the United States, nulliparous, black, or with a history of preterm birth, stillbirth, cesarean delivery, chronic hypertension, or pregestational diabetes were overrepresented in the group that developed preeclampsia. Median gestational age at time of cfDNA measurement was 12.6 weeks, with n=16,653 (95.1%) patients having cfDNA measurement <20 weeks gestation. cfDNA parameters were also significantly different in the group that developed preeclampsia. In the preeclampsia group, total cfDNA was significantly higher (median 362.3 DNA copies / ml vs. 339.0 DNA copies / ml; p<0.001) and FF was significantly lower (median 7.5% vs. 9.4%; p<0.001) (Table 2).

[0071] [Table 2-1] [Table 2-2] [Table 2-3]

[0072] Model Selection and Performance: Exemplary embodiments analyzed and optimized a relatively large number of dense feed-forward neural network (NLNN) models using the protocol described above. More specifically, NLNN models with the number of hidden layers varying between 1 and 5 and the number of neurons varying between 2 and 1024 were considered. In these NLNN models, exemplary embodiments tested different regularization strategies such as L2 kernel regularization, dropout layers, and noise injection in the training set. In some exemplary embodiments, the NLNN models performed similarly to linear neural network models.

[0073] In an example implementation, the NLNN model may be a neural network with one hidden layer composed of 15 reLU units (resulting in a hidden layer with the same dimensions as the input) and L2 regularization implemented as kernel_regularizer in keras.

[0074] The predicted AUC scores for each outcome and the sensitivity for fixed screening positivity rates of 10%, 15% and 20% are shown in Table 3. The numbers shown are average values ​​obtained from the optimization protocol described in the methods and represent the expected behavior of the model for unseen data. The LNN and NLNN model classifiers had similar predictive performance across each of the preeclampsia outcomes (early, early and late preeclampsia). The AUC scores for early, early and late preeclampsia were 0.80, 0.79 and 0.71, respectively, for the NLNN model and 0.80, 0.78 and 0.71, respectively, for the LNN model (Table 3). Sensitivity was highest for the primary outcome, i.e. early preeclampsia with the NLNN model predicting 59% of positive cases and the LNN model predicting 58% of positive cases at a screening positivity rate of 15% (Table 3). In certain embodiments, no difference in predictive performance was observed between the LNN and NLNN models for secondary outcomes (Tables 4 and 5).

[0075] [Table 3]

[0076] [Table 4]

[0077] [Table 5]

[0078] To evaluate the contribution of cfDNA features to the performance of the exemplary models in various embodiments, additional training-validation runs were performed on reduced versions of the dataset in which FF and total cfDNA features were omitted. In these neural network models, the number of units in the hidden layer was rescaled to 13 to match the dimension of the input layer. In an embodiment, for early-onset preeclampsia, removal of total cfDNA and FF features from the model was associated with an approximately 7% decrease in sensitivity at a 15% screening positivity rate for both the LNN and NLNN models (Table 6).

[0079] [Table 6]

[0080] The contribution of ethnicity-related information was assessed through an additional set of models (see Methods for details) to estimate the AUC scores and their 95% confidence intervals in an exemplary embodiment (results are listed in Table 7). After comparing these results with those listed in Tables 3, 4, and 5 (which do not consider ethnicity), it appears that ethnicity-related information does not contribute significantly to the predictive performance of the models.

[0081] [Table 7]

[0082] Feature Importance: In an exemplary embodiment, referring to FIG. 6 and FIG. 7, the input features that were major contributors to the predictions of both the LNN and NLNN models were maternal weight, chronic hypertension, pre-gestational diabetes, previous preterm birth, and previous stillbirth. Parity and increased height showed strong negative correlation with early pre-eclampsia in both models. When cfDNA features were omitted from the model, the remaining 13 features were adapted to compensate for their absence (FIG. 7). In particular, for early-onset pre-eclampsia, not only was a larger relative contribution from IVF features (in vitro fertilization) observed, but also a smaller relative contribution from predictive features related to pre-gestational diabetes (BDIABTYP_PRE) after the omission of cfDNA input variables. It is worth mentioning that the use of matrix multiplication (see Methods section) to evaluate the contribution of input variables of the NLNN model in the presence of reLu hidden units is justified by the fact that in the exemplary optimized model, all arguments of the "hidden" activation functions in the optimized model are non-negative. Figure 7 shows a similar bar plot constructed based on the results obtained from the reduced model in which cfDNA features were omitted, showing how the relative contributions of the remaining 13 features "fit" to compensate for the absence of the missing cfDNA features. The same plot for late-onset preeclampsia shows a similar trend, but for early-onset preeclampsia outcomes, a larger relative contribution from IVF features (in vitro fertilization) and a smaller relative contribution from predictive features related to pregestational diabetes (BDIABTYP_PRE) were observed.

[0083] The exemplary embodiments discussed above show that, in various embodiments, both linear and nonlinear neural network models using patient characteristics collected at the first prenatal visit and features available from SNP-based cfDNA screening (total cfDNA and FFs) can be used to predict early preeclampsia with good levels of performance (expected AUC score = 0.80, 58% sensitivity for a 15% screen positivity rate). The ability of these exemplary models to use only routinely collected clinical data, primarily binary in nature, has practical and economic advantages for clinical implementation.

[0084] In various alternative embodiments, mean arterial pressure may significantly increase the predictive performance of early preeclampsia when added to the model in addition to the other features discussed or in place of one or more of the features discussed. Although blood pressure was not collected as part of SMART, it may be collected early in pregnancy as part of routine clinical practice and incorporated into alternative embodiments of the disclosed model.

[0085] In the disclosed analysis, both cfDNA features (total cfDNA and FF) are associated with the risk of preeclampsia.In various embodiments, FF and cfDNA contribute specifically to the model performance of early-onset preeclampsia.This finding of the performance of biomarkers for early-onset disease may reflect the different pathogenic processes underlying early-onset and late-onset preeclampsia.

[0086] Advantageously, exemplary embodiments of the disclosed approach develop predictive models based solely on routinely collected clinical and cfDNA data, easily obtained from non-invasive prenatal screening, all of which are part of routine clinical care in early pregnancy in the United States. By not requiring other biomarkers (e.g., maternal serum biomarkers) and ultrasound (e.g., uterine artery Doppler), embodiments of the disclosed approach have the advantage of maximizing routinely collected input features without adding the cost of biomarkers and ultrasound that are not part of the current clinical pathway for pregnancy care in the United States. Thus, the disclosed approach can be incorporated into clinical care with minimal resource usage and patient inconvenience.

[0087] Although the present disclosure has been described with respect to specific embodiments, those skilled in the art will recognize that numerous variations are possible. For example, specific examples of rules (including trigger conditions and / or resulting actions) and processes for generating suggested rules are described, but other rules and processes can be implemented. Embodiments of the present disclosure can be implemented using a variety of computer systems and communication technologies, including, but not limited to, the specific specific examples described herein.

[0088] The embodiments of the present disclosure may be realized using any combination of dedicated components and / or programmable processors and / or other programmable devices. The various processes described herein may be implemented on the same processor or different processors in any combination. Where a component is described as being configured to perform a certain operation, such configuration may be achieved, for example, by designing an electronic circuit to perform the operation, by programming a programmable electronic circuit (such as a microprocessor) to perform the operation, or by any combination thereof. Furthermore, while the embodiments described above may refer to specific hardware and software components, those skilled in the art will understand that different combinations of hardware and / or software components may also be used, and that certain operations described as being implemented in hardware may also be implemented in software, and vice versa.

[0089] A computer program incorporating various features of the present disclosure may be coded and stored on various computer readable storage media, suitable media include magnetic disks or tapes, optical storage media such as compact disks (CDs) or DVDs (digital versatile disks), flash memory, and other non-transitory media. A computer readable medium coded with the program code may be packaged with a compatible electronic device, or the program code may be provided separately from the electronic device (e.g., via internet download or as a separately packaged computer readable storage medium).

[0090] Therefore, although the disclosure has been described in terms of specific embodiments, it will be understood that the disclosure is intended to cover all modifications and equivalents that fall within the scope of the following claims.

[0091] As used herein, a "routine" prenatal visit to a health care provider is a "check-up" type visit to a health care provider for the purpose of maintaining health and wellness. Routine visits are intended to prevent adverse health outcomes, not treat existing symptoms or illnesses. Routine prenatal visits are generally recommended for the population of pregnant individuals, as opposed to being prompted by a patient-specific health concern.

[0092] The following include possible embodiments of the disclosed approaches, which represent other examples that are not intended to be limiting in any way.

[0093] Embodiment A1: A method of applying, by a computing system, a predictive machine learning classifier to generate a prediction of a patient who will develop a preeclampsia condition during a current pregnancy, wherein the predictive machine learning classifier comprises one or more artificial neural networks, and generating the prediction comprises: obtaining patient data by at least one of: (A) performing a non-invasive prenatal screening (NIPS) on the patient; or (B) receiving data from a health record system by the computing system over a communications network, the patient data comprising: (i) a set of health characteristics; and (ii) one or more cell-free DNA (cfDNA) metrics; providing the patient data to the predictive machine learning classifier to obtain a prediction of the patient who will develop the preeclampsia condition during the current pregnancy; and applying the predictive machine learning classifier to the computing system. and providing the prediction for care of the patient, by a computing system, wherein providing the prediction comprises at least one of transmitting the prediction by the computing system via the communications network to at least one of the health record system or a computing device associated with a healthcare provider, wherein the predictive machine learning classifier has been trained by applying one or more deep learning techniques to data regarding subjects in a cohort of pregnant subjects, the data including, for each pregnant subject, (A) a first feature set indicative of one or more outcomes of one or more previous pregnancies of the pregnant subject, (B) a second feature set including one or more health indicators for the pregnant subject, and (C) a third feature set based on one or more cfDNA measurements of the pregnant subject.

[0094] Embodiment A2: The method of embodiment A1, wherein all of the patient data used to obtain the prediction is collected non-invasively during one or more routine medical visits of the patient.

[0095] Embodiment A3: The method of any of embodiments A1 or A3, wherein the patient data used to obtain the prediction does not include ultrasound data.

[0096] Embodiment A4: The method of any of embodiments A1 to A3, wherein said preeclampsia condition is early-onset premature preeclampsia.

[0097] Embodiment A5: A method according to any of embodiments A1 to A4, further comprising generating, by the computing system, a training dataset based on the first, second and third feature sets.

[0098] Embodiment A6: The method of any of embodiments A1 to A5, further comprising training, by the computing system, the predictive machine learning classifier using the data regarding the subjects in the cohort of pregnant subjects.

[0099] Embodiment A7: The method of any of embodiments A1 to A6, wherein the predictive machine learning classifier comprises one or more rectified linear units (reLu).

[0100] Embodiment A8: A method according to any of embodiments A1 to A7, wherein the predictive machine learning classifier includes a hidden layer having the same dimensions as the input layer.

[0101] Embodiment A9: A method according to any of embodiments A1 to A8, wherein the one or more deep learning techniques include using one or more regularizers.

[0102] Embodiment B1: A method comprising generating a predictive machine learning classifier for preeclampsia status in a pregnant patient, wherein the predictive machine learning classifier comprises one or more artificial neural networks, and generating the predictive machine learning classifier comprises generating, by a computing system, a training dataset using data for subjects in a cohort of pregnant subjects, the training dataset comprising, for each pregnant subject, (A) a first feature set indicative of one or more outcomes of one or more previous pregnancies of the subject, (B) a second feature set comprising one or more health indicators for the subject, and (C) a third feature set based on one or more cfDNA metrics; and applying, by the computing system, one or more deep learning techniques to the training dataset to generate the predictive machine learning classifier. and providing, by the computing system, the predictive machine learning classifier to generate a patient-specific index for the individual patient, wherein providing the predictive machine learning classifier comprises at least one of: (i) transmitting, by the computing system via a communications network, the predictive machine learning classifier to a second computing system; or (ii) storing the predictive machine learning classifier in a non-transitory computer-readable storage medium accessible for subsequent use by a healthcare provider.

[0103] Embodiment B2: The method of embodiment B1, wherein the predictive machine learning classifier comprises one or more hidden rectified linear units (reLu).

[0104] Embodiment B3: The method of any of embodiments B1 or B2, wherein the predictive machine learning classifier includes a hidden layer having the same dimensions as the input layer.

[0105] Embodiment B4: The method of any of embodiments B1 to B3, wherein applying the one or more deep learning techniques includes using one or more regularizers.

[0106] Embodiment B5: The method of any of embodiments B1 to B4, wherein the one or more regularizers include an L2 regularizer.

[0107] Embodiment B6: The method of any of embodiments B1-B5, further comprising evaluating the patient during a current pregnancy by performing a non-invasive prenatal screening (NIPS) on the patient using the predictive machine learning classifier, and feeding data to the predictive machine learning classifier to obtain a prediction of development of the preeclampsia condition during the current pregnancy, wherein the data includes data from the NIPS, and using the prediction in delivering medical care to the patient, wherein all of the data provided to the predictive machine learning classifier to obtain the prediction is collected non-invasively during one or more regular medical visits for the patient.

[0108] Embodiment B7: The method of any of embodiments B1 to B6, further comprising using the predictive machine learning classifier to evaluate the patient during the current pregnancy by receiving data corresponding to one or more previous pregnancies, one or more health characteristics, and one or more cfDNA metrics of the patient from a health record system, and feeding the data to the predictive machine learning classifier to obtain a prediction of the development of the preeclampsia condition during the current pregnancy, and using the prediction in the delivery of medical care to the patient, wherein all of the data input into the predictive machine learning classifier to obtain the prediction is collected non-invasively during one or more routine medical visits of the patient.

[0109] Embodiment B8: The method of any of embodiments B1 to B7, wherein said preeclampsia condition is early-onset premature preeclampsia.

[0110] Embodiment C1: A computing system configured to apply a predictive machine learning classifier to generate a prediction of a patient who will develop a preeclampsia condition during a current pregnancy, the predictive machine learning classifier comprising one or more artificial neural networks, the computing system comprising one or more processors configured to receive, via a communications network, patient data during the patient's current pregnancy, the patient data comprising a set of health characteristics and one or more cfDNA measurements, provide the patient data to the predictive machine learning classifier to obtain a prediction of the patient who will develop the preeclampsia condition during the current pregnancy, and provide the prediction for care of the patient ... and transmitting the prediction via the communications network to at least one of a computing device associated with a health record system or a healthcare provider, wherein the predictive machine learning classifier has been trained by applying one or more deep learning techniques to data regarding subjects in a cohort of pregnant subjects, and the training data includes, for each pregnant subject, (A) a first feature set indicative of one or more outcomes of one or more previous pregnancies of the pregnant subject, (B) a second feature set corresponding to one or more health indicators of the pregnant subject, and (C) a third feature set corresponding to one or more cell-free DNA metrics for the pregnant subject.

[0111] Embodiment C2: A computing system as described in embodiment C1, wherein all of the patient data used to obtain the prediction is collected non-invasively during one or more routine medical visits of the patient.

[0112] Embodiment C3: A computing system described in any of embodiments C1 or C2, wherein the preeclampsia condition is early-onset premature preeclampsia.

[0113] Embodiment C4: A computing system described in any of embodiments C1 to C3, wherein the one or more processors are further configured to generate the training dataset and apply the deep learning technique to the training dataset to generate the predictive machine learning classifier.

[0114] Embodiment C5: A computing system described in any of embodiments C1 to C4, wherein the one or more processors are further configured to receive at least a subset of the health data from the health record system via the communications network.

[0115] Embodiment C6: A computing system described in any of embodiments C1 to C5, wherein the one or more processors are further configured to receive at least a subset of the health data via a software application executed on the computing device associated with the healthcare provider.

[0116] Embodiment D1: A method for developing a machine learning model useful for predicting the onset of early preeclampsia, the method comprising: identifying, by one or more processors, for each subject of a plurality of subjects, data comprising at least three of maternal age, maternal weight, gestational age, total cell-free DNA, and a fetal fraction percentile for a fetal fraction distribution corresponding to the maternal weight and gestational age of the sample, wherein the data further comprises one of a first label indicating that the subject developed early preeclampsia or a second label indicating that the subject did not develop early preeclampsia; and training, by the one or more processors, a machine learning model using the identified data to generate a value indicative of the subject's risk of developing early preeclampsia.

[0117] Embodiment D2: The method of embodiment D1, wherein the machine learning model includes at least one of a regularized logistic regression model or a feedforward neural network.

[0118] Embodiment D3: The method of any of embodiments D1 or D2, wherein the model comprises a binary classifier.

[0119] Embodiment D4. The method of any of embodiments D1-D3, wherein the model is trained and validated using a stratified k-fold cross-validation (SRKFCV) approach.

[0120] Embodiment D5. The method of any of embodiments D1-D4, wherein the machine learning model is trained by: a) initializing a model parameter accumulator and a score accumulator; b) assigning a first portion of a dataset to a training dataset and a second portion of the dataset to a validation dataset; c) oversampling a minority class in the training dataset to generate an oversampled training dataset; d) oversampling a minority class in the validation dataset to generate an oversampled validation dataset; e) fitting the machine learning model to the oversampled training dataset; f) generating a predicted outcome for the oversampled validation dataset; g) determining an accuracy score and an area under the curve (AUC) score based on the predicted outcome; and h) storing in one or more data structures one or more model parameters in the model parameter accumulator and the accuracy scores and the AUC scores in the score accumulator.

[0121] Embodiment D6. The method of embodiment D5, wherein steps b) through h) are repeated a predetermined number of times.

[0122] Embodiment D7. The method of embodiment D5, wherein the machine learning model is logistic regression and the regularization constant is a hyperparameter.

[0123] Embodiment D8. The method of embodiment D5, wherein the machine learning model is a feedforward neural network and a regularization constant, activation function, number and size of hidden layers, and cost function are optimized.

[0124] Embodiment E1. A method for predicting the onset of early preeclampsia comprising: identifying, by one or more processors, input data for a subject, the input data including at least three of maternal age, maternal weight, gestational age, total cell-free DNA, and a fetal fraction percentile for a fetal fraction distribution corresponding to the maternal weight and gestational age of the sample; determining, by the one or more processors, a value indicative of the subject's risk of developing early preeclampsia in response to the one or more processors inputting the input data into a machine learning model trained using the data to generate a value indicative of the subject's risk of developing early preeclampsia; and providing, by the one or more processors, an output for presentation based on the value.

[0125] Embodiment E2: The method of embodiment E1, wherein the machine learning model comprises at least one of a regularized logistic regression model or a feedforward neural network.

[0126] Embodiment E3: The method of any of embodiments E1 or E2, wherein the model comprises a binary classifier.

[0127] Embodiment E4: The method of any of embodiments E1 to E3, wherein the model is trained and validated using a stratified k-fold cross-validation (SRKFCV) approach.

[0128] Embodiment E5: The method of any of embodiments E1-E4, wherein the machine learning model is trained by: a) initializing a model parameter accumulator and a score accumulator; b) assigning a first portion of a dataset to a training dataset and a second portion of the dataset to a validation dataset; c) oversampling a minority class in the training dataset to generate an oversampled training dataset; d) oversampling a minority class in the validation dataset to generate an oversampled validation dataset; e) fitting the machine learning model to the oversampled training dataset; f) generating a predicted outcome for the oversampled validation dataset; g) determining an accuracy score and an area under the curve (AUC) score based on the predicted outcome; and h) storing in one or more data structures one or more model parameters in the model parameter accumulator and the accuracy scores and the AUC scores in the score accumulator.

[0129] Embodiment E6: The method of embodiment E5, wherein steps b) through h) are repeated a predetermined number of times.

[0130] Embodiment E7: The method of embodiment E5, wherein the machine learning model is logistic regression and the regularization constant is a hyperparameter.

[0131] Embodiment E8: The method of embodiment E5, wherein the machine learning model is a feedforward neural network and a regularization constant, an activation function, the number and size of hidden layers, and a cost function are optimized.

[0132] Embodiment F: A computing system comprising one or more processors configured to implement any of embodiments A1-B8 or D1-E8.

[0133] Embodiment E: A non-transitory computer-readable storage medium having instructions configured to cause one or more processors of a computing system to implement any of embodiments A1-B8 or D1-E8.

[0134] The present technology is not limited with respect to the specific embodiments described in this application, and these embodiments are intended as single illustrations of individual aspects of the present technology. As will be apparent to those skilled in the art, many modifications and variations of the present technology can be made without departing from its spirit and scope. In addition to those listed herein, functionally equivalent methods and devices within the scope of the present technology will be apparent to those skilled in the art from the foregoing description. Such modifications and variations are intended to fall within the scope of the present technology. It will be understood that the present technology is not limited to specific methods, reagents, compound compositions, or biological systems, which may, of course, vary. It will also be understood that the terminology used herein is for the purpose of describing specific embodiments only, and is not intended to be limiting.

[0135] Additionally, when features or aspects of the disclosure are described in terms of a Markush group, those skilled in the art will recognize that the disclosure is thereby also described in terms of any individual member or subgroup of members of the Markush group.

[0136] As would be understood by one of ordinary skill in the art, for any or all purposes, especially in terms of providing a written description, all ranges disclosed herein also encompass any and all possible subranges and combinations of those subranges. Any recited range can be readily recognized as fully descriptive and allowing for the same range to be broken down into at least equal halves, thirds, quarters, fifths, tenths, etc. As a non-limiting example, each range discussed herein can be readily broken down into a lower third, middle third, and upper third, etc. As would also be understood by one of ordinary skill in the art, all language such as "up to," "at least," "greater than," "less than," etc. refers to a range that includes the recited numbers and can then be broken down into subranges as discussed above. Finally, as would be understood by one of ordinary skill in the art, a range includes each individual member. Thus, for example, a group having 1-3 cells refers to a group having 1, 2, or 3 cells. Similarly, a group having 1-5 cells refers to a group having 1, 2, 3, 4, or 5 cells, etc.

[0137] All patents, patent applications, provisional applications, and publications mentioned or cited in this specification are incorporated by reference in their entirety, including all figures and tables, to the extent they do not contradict the explicit teachings of this specification.

Claims

1. 1. A method of applying, by a computing system, a predictive machine learning classifier to generate a prediction of a patient who will develop early preeclampsia during a current pregnancy, the predictive machine learning classifier comprising one or more artificial neural networks, and generating the prediction comprises: (A) obtaining patient data by at least one of: (A) administering a non-invasive prenatal screening (NIPS) to the patient; or (B) receiving data from a health record system by the computing system via a communications network, the patient data including (i) a set of health characteristics, and (ii) one or more cell-free DNA (cfDNA) metrics; feeding the patient data to the predictive machine learning classifier to obtain a prediction that the patient will develop early preeclampsia during the current pregnancy; providing, by the computing system, the prediction for care of the patient, wherein providing the prediction includes at least one of transmitting, by the computing system over the communications network, the prediction to at least one of a computing device associated with the health record system or a healthcare provider; The method of claim 1, wherein the predictive machine learning classifier is trained by applying one or more deep learning techniques to data regarding subjects in a cohort of pregnant subjects, the data comprising, for each pregnant subject, (A) a first feature set indicative of one or more outcomes of the pregnant subject's one or more previous pregnancies, (B) a second feature set comprising one or more health indicators for the pregnant subject, and (C) a third feature set based on one or more cfDNA measurements of the pregnant subject.

2. 2. The method of claim 1, wherein all of the patient data used to derive the prediction (i) is collected non-invasively during one or more routine medical visits of the patient, and (ii) does not include ultrasound data.

3. 2. The method of claim 1, wherein the pre-eclampsia is early-onset pre-eclampsia.

4. The method of claim 1 , further comprising generating, by the computing system, a training data set based on the first, second and third feature sets.

5. 10. The method of claim 1, further comprising training, by the computing system, the predictive machine learning classifier using the data regarding the subjects in the cohort of pregnant subjects.

6. 2. The method of claim 1 , wherein the predictive machine learning classifier includes (i) one or more rectified linear units (reLu) and (ii) a hidden layer having the same dimensions as an input layer, and the one or more deep learning techniques include using one or more regularizers.

7. 1. A method, comprising: generating a predictive machine learning classifier for early preeclampsia in a pregnant patient, the predictive machine learning classifier comprising one or more artificial neural networks, and generating the predictive machine learning classifier comprises: generating, by a computing system, a training dataset using data for subjects in a cohort of pregnant subjects, the training dataset comprising, for each pregnant subject: (A) a first set of features indicative of one or more outcomes of one or more prior pregnancies of the subject; (B) a second set of features corresponding to one or more health indicators of the subject; and (C) generating a training dataset including a third feature set corresponding to one or more cfDNA metrics; applying, by the computing system, one or more deep learning techniques to the training dataset to generate the predictive machine learning classifier, the predictive machine learning classifier configured to receive as inputs patient health characteristics and one or more cfDNA metrics and to output a prediction regarding the occurrence of early preeclampsia; and providing, by the computing system, the predictive machine learning classifier to generate a patient-specific index for an individual patient, wherein providing the predictive machine learning classifier includes at least one of: (i) transmitting, by the computing system over a communications network, the predictive machine learning classifier to a second computing system, or (ii) storing the predictive machine learning classifier in a non-transitory computer-readable storage medium accessible for subsequent use by a healthcare provider.

8. The method of claim 7 , wherein the predictive machine learning classifier comprises one or more hidden rectified linear units (reLu).

9. The method of claim 7 , wherein the predictive machine learning classifier includes a hidden layer having the same dimensions as an input layer.

10. The method of claim 7 , wherein applying the one or more deep learning techniques comprises using one or more regularizers.

11. The method of claim 10 , wherein the one or more regularizers include an L2 regularizer.

12. Using the predictive machine learning classifier, administering non-invasive prenatal screening (NIPS) to said patient; providing data to the predictive machine learning classifier to obtain a prediction of the occurrence of the early preeclampsia in a current pregnancy, the data comprising data from the NIPS; and using the prediction in the delivery of medical care to the patient, wherein all of the data provided to the predictive machine learning classifier to derive the prediction is collected non-invasively during one or more routine medical visits of the patient; 8. The method of claim 7, further comprising evaluating the patient during the current pregnancy by

13. Using the predictive machine learning classifier, receiving data from a health record system corresponding to one or more previous pregnancies, one or more health characteristics, and one or more cfDNA metrics of the patient; feeding the data to the predictive machine learning classifier to obtain a prediction of the occurrence of early preeclampsia in a current pregnancy; using the prediction in the delivery of medical care to the patient; 8. The method of claim 7, further comprising using the predictive machine learning classifier to evaluate a patient during the current pregnancy by:

14. 2. The method of claim 1, wherein the pre-eclampsia is early-onset pre-eclampsia.

15. 1. A computing system configured to apply a predictive machine learning classifier to generate a prediction of a patient who will develop early preeclampsia during a current pregnancy, the predictive machine learning classifier comprising one or more artificial neural networks, the computing system comprising: receiving, via a communications network, patient data during the current pregnancy of the patient, the patient data including a set of health characteristics and one or more cfDNA measurements; feeding the patient data to the predictive machine learning classifier to obtain a prediction that the patient will develop the early preeclampsia during the current pregnancy; one or more processors configured to provide the prediction for care of the patient, wherein providing the prediction includes transmitting the prediction over the communications network to at least one of a health record system or a computing device associated with a health care provider; 1. A computing system comprising: a computing device for detecting and predicting a cell-free DNA (C) population of subjects in a cohort of pregnant subjects, the computing device comprising: a first feature set indicative of one or more outcomes of one or more previous pregnancies of the pregnant subject; a second feature set corresponding to one or more health indicators of the pregnant subject; and a third feature set corresponding to one or more cell-free DNA metrics for the pregnant subject.

16. 20. The computing system of claim 15, wherein all of the patient data used to derive the prediction is collected non-invasively during one or more routine medical visits of the patient.

17. 16. The computing system of claim 15, wherein the pre-eclampsia is early-onset pre-eclampsia.

18. 16. The computing system of claim 15, wherein the one or more processors are further configured to generate the set of training data and apply the deep learning technique to the set of training data to generate the predictive machine learning classifier.

19. 16. The computing system of claim 15, wherein the one or more processors are further configured to receive at least a subset of health data from the health record system via the communications network.

20. 16. The computing system of claim 15, wherein the one or more processors are further configured to receive at least a subset of the health data via a software application executing on the computing device associated with the health care provider.

Images