Methods and systems for classifying the occurrence of medical complications

Abstract

This invention relates to a computer-implemented method for classifying the occurrence of complications caused by preeclampsia within a predetermined time period, comprising: applying (S3) a second machine learning algorithm (A2) to a second dataset (DS2) of medical parameters of a patient to classify the occurrence of complications caused by preeclampsia within the predetermined time period; and outputting (S4) a class representing the occurrence of complications caused by preeclampsia within the predetermined time period. The invention also relates to a system (1) for classifying the occurrence of complications caused by preeclampsia within a predetermined time period.

Description

Technical Field

[0001] The present invention relates to a computer-implemented method for classifying the occurrence of complications caused by preeclampsia within a predetermined time period.

[0002] The present invention also relates to a system for classifying the occurrence of complications caused by preeclampsia within a predetermined time period.

[0003] Furthermore, the present invention relates to a computer program having program code for performing the method of the present invention, and a computer-readable medium having program code of the computer program, wherein when the computer program is executed on a computer, the computer program is used to perform at least a portion of the method of the present invention. Background Technology

[0004] Preeclampsia is a multisystemic disease of pregnancy. With an incidence of 3-5%, it is one of the leading causes of maternal and fetal morbidity and mortality. Preeclampsia is defined as elevated blood pressure during pregnancy, accompanied by new-onset maternal organ symptoms such as proteinuria, liver dysfunction, and neurological symptoms. Currently, the prediction of preeclampsia is based on effective national and international guidelines, through regular measurement of maternal blood pressure and proteinuria.

[0005] However, these diagnostic criteria are too imprecise to predict the severe course or complications of preeclampsia, posing a clinical dilemma. Accurate prediction is crucial for preventing these diseases, given the variability between nonspecific symptoms and life-threatening complications.

[0006] With the establishment of angiogenesis and anti-angiogenic factors, the prediction of preeclampsia has been improved. It has been shown that the circulating serum concentrations of the anti-angiogenic factor soluble FMS-like tyrosine kinase 1 (sFlt-1) and the angiogenic factor placental growth factor (P1GF) are altered in the serum of women who are likely to develop preeclampsia.

[0007] In “Predictive Value of the sFlt-1:PlGF Ratio in Women with Suspected Preeclamp S1a”, Zeisler H, Llurba E, Chantraine F, Vatish M, Staff AC, Sennström M et al., N Engl J Med. 2016;374(1):13-22, it was demonstrated that the sFlt-1 / PlGF ratio could be used to exclude women with clinical symptoms. Due to the moderate positive predictive value, inclusion was limited.

[0008] Therefore, high predictive accuracy is needed to reduce morbidity and mortality rates and avoid unnecessary hospitalizations, thereby alleviating the burden on the healthcare system.

[0009] Therefore, the object of the present invention is to provide an improved method that enables more accurate determination of the occurrence of complications caused by preeclampsia during pregnancy. Summary of the Invention

[0010] According to the present invention, this objective is achieved by a computer-implemented method having the features of claim 1 for classifying the occurrence of complications caused by preeclampsia within a predetermined time period.

[0011] According to the invention, this objective is further achieved by a system having the features of claim 11 for classifying the occurrence of complications caused by preeclampsia within a predetermined time period. According to the invention, this objective is also achieved by a computer program having program code for executing the method of the invention, having the features of claim 14.

[0012] Furthermore, the objective is achieved through a computer-readable medium having program code of a computer program, so that at least a portion of the method of the invention is performed when the computer program is executed on a computer, having the features of claim 15.

[0013] The present invention relates to a computer-implemented method for classifying the occurrence of complications caused by preeclampsia within a predetermined time period.

[0014] The method includes providing a pre-recorded first dataset of the patient's medical parameters.

[0015] Furthermore, the method includes using a first algorithm, particularly a multivariate data analysis method, to at least partially transform a first dataset of patients' medical parameters into a simplified second dataset of patients' medical parameters.

[0016] The method further includes applying a second machine learning algorithm to the second dataset of the patient's medical parameters to classify the occurrence of the complications caused by preeclampsia within the predetermined time period, and outputting a classification representing the occurrence of the complications caused by preeclampsia within the predetermined time period.

[0017] The present invention also relates to a system for classifying the occurrence of complications caused by preeclampsia within a predetermined time period. The system includes at least one input device for providing a pre-recorded first dataset of the patient's medical parameters.

[0018] In addition, the system includes a first computing device configured to use a first algorithm, particularly a multivariate data analysis method, to at least partially convert a first dataset of the patient's medical parameters into a simplified second dataset of the patient's medical parameters.

[0019] The system also includes a second computing device configured to apply a second machine learning algorithm to a second dataset of the patient’s medical parameters to classify the occurrence of complications caused by preeclampsia within a predetermined time period, and an output device configured to output a classification representing the occurrence of complications caused by preeclampsia within the predetermined time period.

[0020] The present invention also relates to a computer program having program code that, when executed on a computer, executes a method for classifying the occurrence of complications caused by preeclampsia within a predetermined time period, and a computer-readable medium having program code that, when executed on a computer, executes at least a portion of the method of the present invention.

[0021] Machine learning algorithms are based on using statistical methods to train data processing systems to perform specific tasks without initially requiring explicit programming. The goal of machine learning is to build algorithms that can learn from data and make predictions. These algorithms create mathematical models that can, for example, classify data.

[0022] A first algorithm, particularly a multivariate data analysis method, performs feature selection and feature processing to transform a first dataset of patient medical parameters into a simplified second dataset of patient medical parameters. Relevant features are selected from said dataset and transformed into a format usable by a second machine learning algorithm. This may include feature extraction, feature scaling, and feature normalization.

[0023] One approach of this invention is to apply machine learning algorithms to the field of risk prediction for pregnant women at high risk of preeclampsia. In cases where potential complications are predicted, termination of pregnancy may be recommended, for example. This can eliminate the risk of complications. Predictions can be made throughout the pregnancy, preferably after 24 weeks of gestation, or whenever a high-risk case arises.

[0024] Machine learning algorithms are particularly well-suited for simplified datasets of patients' medical parameters. In general, machine learning models function by learning from data, identifying patterns and relationships between variables, and making predictions or decisions based on them. The accuracy of the model depends on the quality and quantity of the data, the algorithm used, and the settings of the hyperparameters.

[0025] Feature selection and engineering are key steps in creating machine learning models because they aim to select the most important features from a dataset and transform them into a format that machine learning algorithms can use.

[0026] During feature selection, a subset of relevant features is chosen for use in the model. The goal of feature selection is to reduce the dimensionality of the data and remove all redundant or irrelevant features that may negatively impact model performance.

[0027] The classification of complications arising from preeclampsia within a predetermined time period can be provided, for example, by a binary classification that distinguishes only between the prediction of complication occurrence and the absence of complication.

[0028] Optionally, the classification can output, for example, the percentage probability of the occurrence of complications, expressed as a value between 0 and 1.

[0029] In addition, a platform was created that integrates physician-side decision support tools with patient-side home monitoring applications.

[0030] Other embodiments of the invention are determined by the appended claims and the following description with reference to the accompanying drawings.

[0031] According to a preferred embodiment of the present invention, the first dataset includes a first set of medical parameters of a patient recorded through a user interface of a medical practitioner, including protein levels in urine, systolic blood pressure, diastolic blood pressure, SFLT-1 value, P1GF value, classification of thrombocytopenia, and classification of elevated liver enzymes.

[0032] Therefore, a portion of the features included in the first dataset can be advantageously provided by medical practitioners through a user interface.

[0033] According to another preferred embodiment of the invention, the first dataset includes a second set of medical parameters of the patient recorded through a patient app and / or a medical practitioner's user interface, including the patient's age, weight, height, hypertension classification before current pregnancy, hypertension classification newly developed during the current pregnancy, and gestational age, particularly in days.

[0034] Therefore, another set of features included in the first dataset can be advantageously provided by the patient through the patient app.

[0035] According to another preferred embodiment of the present invention, a second machine learning algorithm is specified to further classify the occurrence of complications caused by preeclampsia within a predetermined time period based on the re-provided medical parameters of the patient recorded by the patient app.

[0036] Therefore, the risk of complications arising from preeclampsia can be advantageously determined continuously based on updated medical parameters provided by the patient app.

[0037] According to another preferred embodiment of the invention, the second dataset provided includes the patient's age, pre-pregnancy body mass index, multiples of the median SFLT-1, multiples of the median P1GF, multiples of the median SFLT-L / PIGF ratio, deviation from mean systolic blood pressure, and deviation from mean diastolic blood pressure.

[0038] Therefore, the transformed second dataset advantageously contains relevant features and a format that can be used by a second machine learning algorithm.

[0039] According to another preferred embodiment of the invention, a second machine learning algorithm and / or a third machine learning algorithm, applied to a second dataset of the patient's medical parameters, outputs a classification of labor caused by preeclampsia within a predetermined time period. This prediction advantageously provides an improved basis for decision-making regarding any necessary medical interventions.

[0040] According to another preferred embodiment of the present invention, a second machine learning algorithm and / or a third machine learning algorithm applied to a second dataset of the patient's medical parameters outputs a classification of labor caused by preeclampsia in patients before the 34th week of pregnancy within the next 14 days, and a classification of labor caused by preeclampsia in patients after the 34th week of pregnancy within the next 7 days.

[0041] Based on this certainty, doctors can make better decisions about whether a patient is at high risk and needs hospital treatment, whether the patient should be delivered immediately or soon, or whether, despite symptoms, the patient can be sent home with greater certainty than before.

[0042] According to another preferred embodiment of the present invention, a second machine learning algorithm and / or a fourth machine learning algorithm applied to a second dataset of the patient's medical parameters outputs a classification of the occurrence of preeclampsia.

[0043] Predicting the onset of preeclampsia can advantageously provide physicians with additional information needed for treatment. For example, the occurrence of preeclampsia can be predicted, as well as the absence of complications arising from it. This allows for improved decision-making regarding the patient's potential treatment.

[0044] According to another preferred embodiment of the present invention, a computer-implemented method is provided, which further includes providing a classification generated by a second machine learning algorithm, a third machine learning algorithm and / or a fourth machine learning algorithm.

[0045] The provided classification results can then be transmitted, for example, to the user interface of healthcare practitioners and / or patient apps.

[0046] According to another preferred embodiment of the present invention, the classification of the occurrence of complications caused by preeclampsia within a predetermined time period includes a first classification, which represents the prediction of the occurrence of complications caused by preeclampsia within the predetermined time period, and a second classification, which represents the prediction of the non-occurrence of complications caused by preeclampsia within the predetermined time period.

[0047] Therefore, binary classification allows for specific statements about the occurrence or non-occurrence of complications caused by preeclampsia within a predetermined time period.

[0048] According to another preferred embodiment of the invention, the at least one input device includes a user interface for a medical practitioner, the user interface being configured to provide a first set of medical parameters of the patient recorded by the medical practitioner, the first set of medical parameters including protein levels in urine, systolic blood pressure, diastolic blood pressure, SFLT-1 value, PIGF value, classification of thrombocytopenia, and classification of elevated liver enzymes.

[0049] Therefore, a portion of the features included in the first dataset can be advantageously provided by medical practitioners through a user interface.

[0050] According to another preferred embodiment of the invention, the at least one input device includes a patient app configured to provide a second set of medical parameters recorded by the patient, the second set of medical parameters including the patient's age, weight before current pregnancy, height, hypertension classification before current pregnancy, hypertension classification newly occurring during current pregnancy, and gestational age, particularly in days.

[0051] Therefore, another set of features included in the first dataset can be advantageously provided by the patient through the patient app.

[0052] The features of the computer-implemented method described herein for classifying the occurrence of complications caused by preeclampsia within a predetermined time period are also applicable to systems for classifying the occurrence of complications caused by preeclampsia within a predetermined time period, and vice versa. Attached Figure Description

[0053] Brief Description of the Drawings To better understand the present invention and its advantages, reference is now made to the following description in conjunction with the accompanying drawings.

[0054] The invention will now be explained in more detail by way of exemplary embodiments indicated in the schematic illustrations of the accompanying drawings.

[0055] In the attached diagram:

[0056] Figure 1 A flowchart is shown of a computer-implemented method for classifying the occurrence of complications caused by preeclampsia within a predetermined time period; and

[0057] Figure 2 A schematic diagram of a system for classifying the occurrence of complications caused by preeclampsia within a predetermined time period is shown.

[0058] Unless otherwise stated, the same reference numerals denote the same elements in the drawings. Detailed Implementation

[0059] Figure 1 The computer-implemented method shown for classifying the occurrence of complications caused by preeclampsia within a predetermined time period includes providing a first dataset DS1 of pre-recorded medical parameters of patients S1.

[0060] Furthermore, the method includes using a first algorithm A1, particularly a multivariate data analysis method, to at least partially transform a first dataset DS1 of the patient's medical parameters into a simplified second dataset DS2 of the patient's medical parameters.

[0061] The method also includes applying a second machine learning algorithm A2 to a second dataset DS2 of the patient's medical parameters to classify the occurrence of complications caused by preeclampsia within a predetermined time period.

[0062] In addition, the method includes output S4 representing the categories K1 and K2 of complications caused by preeclampsia occurring within a predetermined time period.

[0063] The first dataset DS1 includes a first set of medical parameters of patients recorded via a medical practitioner's user interface 10, including protein levels in urine, systolic blood pressure, diastolic blood pressure, SFLT-1 value, P1GF value, classification of thrombocytopenia, and classification of elevated liver enzymes.

[0064] The first dataset DS1 also includes a second set of medical parameters of the patient recorded via the patient app 12 and / or the user interface 10 of the medical practitioner, including the patient's age, weight, height before current pregnancy, hypertension classification before current pregnancy, hypertension classification newly developed during the current pregnancy, and gestational age, specifically in days.

[0065] The second machine learning algorithm A2 performs further classification of the occurrence of complications caused by preeclampsia within a predetermined time period based on the re-providalization of the patient's medical parameters recorded via the patient app 12.

[0066] The transformed second dataset DS2 includes the patient's age, pre-pregnancy body mass index, multiples of the median SFLT-1, multiples of the median P1GF, multiples of the median SFLT-1 / P1GF ratio, deviation from mean systolic blood pressure, and deviation from mean diastolic blood pressure.

[0067] The second machine learning algorithm A2 and / or the third machine learning algorithm A3, applied to the second dataset DS2 of the patient's medical parameters, further outputs a classification of labor caused by preeclampsia within a predetermined time period. Furthermore, the second machine learning algorithm A2 and / or the third machine learning algorithm A3, applied to the second dataset DS2 of the patient's medical parameters, outputs a classification of labor caused by preeclampsia in the next 14 days for patients before the 34th week of pregnancy, and a classification of labor caused by preeclampsia in the next 7 days for patients after the 34th week of pregnancy. The second machine learning algorithm A2 and / or the fourth machine learning algorithm A4, applied to the second dataset DS2 of the patient's medical parameters, further outputs a classification of the occurrence of preeclampsia.

[0068] In addition, the method includes providing classifications generated by a second machine learning algorithm A2, a third machine learning algorithm A3, and / or a fourth machine learning algorithm A4.

[0069] The classification of the occurrence of complications caused by preeclampsia within a predetermined time period includes Category I K1, which represents the prediction of the occurrence of complications caused by preeclampsia within a predetermined time period.

[0070] In addition, the classification includes a second category, K2, which indicates a prediction that no complications caused by preeclampsia will occur within a predetermined time period.

[0071] Optionally, the classification can output, for example, the percentage probability of the occurrence of complications, expressed as a value between 0 and 1.

[0072] Figure 2 A schematic diagram of System 1 is shown for classifying the occurrence of complications caused by preeclampsia within a predetermined time period. System 1 includes at least one input device 14 for providing a pre-recorded first dataset DS1 of the patient's medical parameters.

[0073] In addition, system 1 includes a first computing device 16 configured to use a first algorithm A1, particularly a multivariate data analysis method, to at least partially convert a first dataset of the patient's medical parameters into a simplified second dataset DS2 of the patient's medical parameters.

[0074] System 1 also includes a second computing device 18 configured to apply a second machine learning algorithm A2 to a second dataset DS2 of the patient’s medical parameters to classify the occurrence of complications caused by preeclampsia within a predetermined time period.

[0075] In addition, system 1 includes an output device 20 configured to output a category of complications caused by preeclampsia occurring within a predetermined time period.

[0076] The at least one input device 14 includes a user interface 10 for a medical practitioner, the user interface being configured to provide a first set of medical parameters of the patient recorded by the medical practitioner, the first set of medical parameters including protein levels in urine, systolic blood pressure, diastolic blood pressure, SFLT-1 value, P1GF value, classification of thrombocytopenia, and classification of elevated liver enzymes.

[0077] In addition, the at least one input device 14 includes a patient app 12, configured to provide a second set of medical parameters recorded by the patient, including the patient's age, weight and height before current pregnancy, hypertension classification before current pregnancy, hypertension classification newly developed during the current pregnancy, and gestational age, particularly in days.

[0078] The user interface 10 for healthcare practitioners and the patient app 12 are configured to communicate with the backend app 22. The backend app 22 is then connected to the model service app 24, where provided medical parameters are transmitted from the backend app 22 to the model service app 24.

[0079] Then, the classification generated by the corresponding machine learning model is transmitted from the model service app 24 back to the backend app 22, and the medical practitioner's user interface 10 and / or the patient app 12 can access the backend app 22.

[0080] The patient app 12 and / or the user interface 10 for healthcare practitioners can be provided, for example, as a desktop app and / or a smartphone app.

[0081] While specific embodiments have been shown and described herein, those skilled in the art will understand that many alternative and / or equivalent implementations exist. It should be noted that one or more exemplary embodiments are merely examples and are not intended to limit the scope, applicability, or configuration in any way.

[0082] Conversely, the foregoing overview and detailed description provide convenient guidance to those skilled in the art for implementing at least one exemplary embodiment. It should be understood that various changes may be made to the function and arrangement of the elements without departing from the scope of the appended claims and their legal equivalents.

[0083] Generally, this application is intended to cover modifications, alterations, or variations of the embodiments presented herein. For example, the order of method steps may be changed. The method may also be performed, at least partially, sequentially or in parallel.

[0084] List of reference numerals

[0085] 1 system

[0086] User interface for 10 healthcare practitioners

[0087] 12 Patients App

[0088] 14 Input Devices

[0089] 16 First Computing Device

[0090] 18 Second Computing Device

[0091] 20 output devices

[0092] 22 backend apps

[0093] 24 Model Service App

[0094] Algorithm 1

[0095] A2 Second Machine Learning Algorithm

[0096] A3 Third Machine Learning Algorithm

[0097] A4 Fourth Machine Learning Algorithm

[0098] DS1 First Dataset

[0099] DS2 Second Dataset

[0100] K1 Class I

[0101] K2 Type II

[0102] S1-S4 Method Steps

Claims

1. A computer-implemented method for classifying the occurrence of complications caused by preeclampsia within a predetermined time period, comprising: Provides a pre-recorded first dataset (DS1) of the patient's medical parameters; Using a first algorithm (A1), particularly a multivariate data analysis method, the first dataset (DS1) of the patient's medical parameters is at least partially transformed (S2) into a simplified second dataset (DS2) of the patient's medical parameters; The second machine learning algorithm (A2) is applied (S3) to the second dataset (DS2) of the patient's medical parameters to classify the occurrence of complications caused by preeclampsia within the predetermined time period; as well as The output (S4) represents the class (K1, K2) of complications caused by preeclampsia that occurred during the predetermined time period.

2. The computer-implemented method according to claim 1, wherein the first dataset (DS1) comprises a first set of medical parameters of the patient recorded by a user interface (10) of a medical practitioner, including protein levels in urine, systolic blood pressure, diastolic blood pressure, sFlt-1 value, P1GF value, classification of thrombocytopenia, and classification of elevated liver enzymes.

3. The computer-implemented method according to claim 2, wherein the first dataset (DS1) includes a second set of medical parameters of the patient recorded by the medical practitioner's patient app (12) and / or the user interface (10), including the patient's age, weight before current pregnancy, height, hypertension classification before current pregnancy, new hypertension classification during current pregnancy, and gestational age, particularly in days.

4. The computer-implemented method according to claim 3, wherein the second machine learning algorithm (A2) further classifies the occurrence of complications caused by preeclampsia within the predetermined time period based on the re-providalization of the patient's medical parameters recorded by the patient app (12).

5. The computer-implemented method according to any one of the preceding claims, wherein the transformed second dataset (DS2) includes the patient's age, pre-pregnancy body mass index, multiples of the median sFlt-1, multiples of the median P1GF, multiples of the median ratio of sFlt-1 and P1GF, deviation from the mean systolic blood pressure, and deviation from the mean diastolic blood pressure.

6. The computer-implemented method according to any one of the preceding claims, wherein a second machine learning algorithm (A2) and / or a third machine learning algorithm (A3) applied to a second dataset (DS2) of medical parameters of the patient outputs a classification of labor caused by preeclampsia within a predetermined time period.

7. The computer-implemented method of claim 6, wherein a second machine learning algorithm (A2) and / or a third machine learning algorithm (A3) applied to a second dataset (DS2) of the patient's medical parameters outputs a classification of labor caused by preeclampsia in the next 14 days for patients before the 34th week of pregnancy, and a classification of labor caused by preeclampsia in the next 7 days for patients after the 34th week of pregnancy.

8. The computer-implemented method according to any one of the preceding claims, wherein a second machine learning algorithm (A2) and / or a fourth machine learning algorithm (A4) applied to a second dataset (DS2) of medical parameters of the patient outputs a classification of the occurrence of preeclampsia.

9. The computer-implemented method of claim 8, further comprising providing the classification generated by the second machine learning algorithm (A2), the third machine learning algorithm (A3), and / or the fourth machine learning algorithm (A4).

10. A computer-implemented method according to any one of the preceding claims, wherein the classification of the occurrence of complications caused by preeclampsia within the predetermined time period includes a first category (K1) representing a prediction of the occurrence of complications caused by preeclampsia within the predetermined time period, and wherein a second category (K2) of the classification represents a prediction of the non-occurrence of complications caused by preeclampsia within the predetermined time period.

11. A system (1) for classifying the occurrence of complications caused by preeclampsia within a predetermined time period, comprising: At least one input device (14) is used to provide a pre-recorded first dataset (DS1) of the patient's medical parameters; A first computing device (16) is configured to use a first algorithm (A1), particularly a multivariate data analysis method, to at least partially transform a first dataset of the patient's medical parameters into a simplified second dataset (DS2) of the patient's medical parameters. The second computing device (18) is configured to apply a second machine learning algorithm (A2) to a second dataset (DS2) of the patient's medical parameters to classify the occurrence of the complications caused by preeclampsia within the predetermined time period; as well as The output device (20) is configured to output a class (K1, K2) representing the occurrence of the complications caused by preeclampsia during the predetermined time period.

12. The system of claim 11, wherein the at least one input device (14) comprises a user interface (10) of a medical practitioner, the user interface being configured to provide a first set of medical parameters of the patient recorded by the medical practitioner, the first set of medical parameters including protein levels in urine, systolic blood pressure, diastolic blood pressure, sFlt-1 value, P1GF value, classification of thrombocytopenia, and classification of elevated liver enzymes.

13. The system according to claim 11 or 12, wherein the at least one input device (14) comprises a patient app (12) configured to provide a second set of medical parameters recorded by the patient, the second set of medical parameters including the patient's age, weight before current pregnancy, height, hypertension classification before current pregnancy, hypertension classification newly occurring during current pregnancy, and gestational age, particularly in days.

14. A computer program having program code, which, when executed on a computer, is used to perform at least a portion of the method according to any one of claims 1 to 10.

15. A computer-readable medium having program code of a computer program, wherein when the computer program is executed on a computer, the computer program is used to perform at least a portion of the method according to any one of claims 1 to 10.

Images