System and method for predicting and managing small-gestational-age fetuses

Abstract

The present invention relates to a system (100) for predicting and managing small gestational age or SGA fetuses. The system (100) is configured to collect a plurality of clinical parameters of a patient and analyze them using a predictive model, wherein the predictive model generates a probabilistic score that predicts the likelihood of a fetus having an SGA condition at birth. The system (100) is further configured to generate a diagnostic report and at least one intervention plan related to the prediction result. The intervention plan enables personalized prenatal care for optimizing their growth and development outcomes when monitoring and managing an SGA fetus. The system (100) comprises a data acquisition module (10), a data encoding module (20), a processing module (30), an inference module (40), a data decoding module (50), a management module (60) and a display module (70).

Description

Technical Field

[0001] The present invention relates to a system and method for predicting and managing a small-for-gestational-age or SGA fetus. More specifically, the present invention relates to a system and method for predicting and managing a SGA fetus using a machine learning algorithm. Background Art

[0002] Small for gestational age or SGA is a term used to describe the condition of a fetus whose birth weight is less than ten percent of its gestational age. A fetus diagnosed with an SGA condition typically has a smaller average weight when compared to fetuses of the same gestational age. These may be due to different factors; an SGA fetus may have experienced growth restriction in the uterus, or be born small due to its genetic makeup. SGA fetuses often require specialized care after birth because of their smaller size and prone to various health complications. Early prediction of the likelihood of the development of an SGA condition in a fetus would enable healthcare providers to implement appropriate interventions to optimize fetal growth and minimize potential complications. Therefore, various systems and methods have been developed to facilitate the detection or prediction of an SGA condition in a fetus.

[0003] An example of a system and method for predicting the SGA status of a fetus is disclosed in European Patent Application Publication No. EP 2641092 A1. The system includes: a determination system for detecting a plurality of metabolite biomarkers in a biological sample of a patient; a storage system for storing metabolite biomarker data generated by the determination system; and a comparison system for comparing the metabolite biomarker data with a metabolite fingerprint to provide a quantitative prediction of SGA. The biological sample is measured for all metabolite biomarkers, wherein the levels of the measured metabolite biomarkers are correlated with the metabolite fingerprint of SGA.

[0004] Although there are various systems and methods for predicting and detecting the SGA condition of a fetus, most systems and methods typically involve invasive and expensive processes, such as collecting and analyzing biological samples from patients. Some systems and methods may also have problems with poor detection or high false negative rates, which means that the SGA condition of a fetus may not be accurately detected before the fetus is delivered. Once the SGA condition is detected in a fetus, these fetuses typically require specialized management or arrangements to meet their unique needs and optimize their growth and development. Therefore, it is necessary to address the above shortcomings. Summary of the invention

[0005] According to one aspect of the present invention, a system (100) for predicting and managing a small for gestational age or SGA fetus is provided. The system (100) comprises: a data acquisition module (10) configured to collect a plurality of clinical parameters; and a processing module (30) configured to perform data preprocessing on the plurality of clinical parameters. The system (100) is characterized in that the system (100) further comprises an inference module (40) configured to analyze the plurality of clinical parameters using a prediction model to predict the risk of an SGA condition in the fetus, wherein the inference module (40) is connected to the processing module (30).

[0006] Preferably, the system (100) further comprises a management module (60) configured to generate a diagnostic report and at least one intervention plan for managing and monitoring a SGA or Appropriate-for-Gestational-Age or AGA fetus.

[0007] Preferably, the system (100) further comprises a display module (70) configured to display data related to the prediction result, the diagnosis report and the at least one intervention plan to a user of the system (100).

[0008] Preferably, the system (100) further comprises a data encoding module (20) configured to execute a data encoding algorithm.

[0009] Preferably, the system (100) further comprises a data decoding module (50) configured to execute a data decoding algorithm.

[0010] Preferably, the inference module (50) quantifies the risk as a probability score and calculates importance scores for a plurality of input features.

[0011] According to another aspect of the present invention, a method for predicting and managing a small for gestational age or SGA fetus is provided. The method is characterized by the following steps: collecting a plurality of clinical parameters by a data acquisition module (10); analyzing the plurality of clinical parameters by an inference module (40) to predict the risk of a SGA or an average for gestational age or AGA condition in the fetus; and generating a diagnostic report and at least one intervention plan related to the prediction of a SGA or AGA condition in the fetus by a management module (60).

[0012] Preferably, the method further comprises: performing a step of data encoding on a plurality of clinical parameters by a data encoding module (20).

[0013] Preferably, the method further comprises: a step of performing data preprocessing on a plurality of clinical parameters by a processing module (30).

[0014] Preferably, the method further comprises: performing a step of data decoding on a plurality of clinical parameters by a data decoding module (50).

[0015] Preferably, the method further comprises the step of displaying data related to the prediction result, the diagnosis report and the at least one intervention plan to a user of the system (100) via a display module (70).

[0016] Preferably, the step of analyzing the plurality of clinical parameters by the inference module (40) to predict the risk of SGA or AGA condition in the fetus further comprises the steps of: assigning the plurality of clinical parameters as input features; inputting the input features into a prediction model; quantifying the risk as a probability score; and calculating an importance score for the input features.

[0017] Preferably, the step of generating at least one intervention plan by the management module (60) further comprises the steps of: analyzing the probability score; if the probability score indicates that the fetus is more likely to be born with SGA condition, analyzing the fetal biometric measurements; analyzing the fetal Doppler measurements; generating an ultrasound recommendation; and generating a delivery recommendation.

[0018] Preferably, if the probability score indicates that the fetus is more likely to be born with an AGA condition, a recommendation is output for a course of action to be taken for caring for the AGA fetus. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments of the invention and, together with the description, serve to explain the principles of the invention.

[0020] Figure 1a A system (100) for predicting and managing a small for gestational age or SGA fetus according to an embodiment of the present invention is shown.

[0021] Figure 1b and Figure 1c An example of a user interface for inputting a plurality of maternal and fetal parameters into a data acquisition module (10) is shown.

[0022] Figure 1d and Figure 1e An example of an interactive dashboard displaying prediction results, diagnostic reports, and at least one intervention plan to a user of the system (100) via a display module (70) is shown.

[0023] Figure 2a A flow chart of a method for predicting and managing a SGA fetus according to an embodiment of the present invention is shown.

[0024] Figure 2b The sub-steps of generating an intervention plan for managing an SGA fetus by the management module (60) are shown, such as Figure 2a The method is shown in .

[0025] Figure 2cAn example of generating an intervention plan by the management module (60) is shown. DETAILED DESCRIPTION

[0026] Preferred embodiments of the present invention will be described herein below with reference to the accompanying drawings. In the following description, well-known functions or constructions are not described in detail since they would obscure the description with unnecessary detail.

[0027] Reference now Figure 1a , which shows a block diagram of a system (100) for predicting and managing a small for gestational age or SGA fetus according to an embodiment of the present invention. The system (100) is configured to collect and analyze multiple clinical parameters of a patient using a predictive model, wherein the predictive model generates a probability score predicting the likelihood of a fetus being born with an SGA condition. In an embodiment of the present invention, the patient is a pregnant woman with a fetus in her uterus. The pregnant woman is preferably in her second trimester, which is 13 to 28 weeks of gestation. The system (100) is also configured to generate a diagnostic report associated with the predicted results and at least one intervention plan. The intervention plan enables personalized prenatal care when monitoring and managing the SGA fetus to optimize its growth and developmental outcomes. The intervention plan allows users of the system (100), such as clinicians and medical staff, to receive insights into a recommended course of action to help deliver a healthy baby.

[0028] The system (100) includes a data acquisition module (10), a data encoding module (20), a processing module (30), an inference module (40), a data decoding module (50), a management module (60), and a display module (70). The modules of the system (100) can be connected via wired communication, wireless communication, or a combination of both. One or more modules of the system (100) can be executed in a remote environment, a cloud environment, or a combination of both. For example, the data acquisition module (10) and the display module (70) can be executed remotely in a clinic, a hospital, or any other medical facility, while the inference module (50) can be executed on a cloud server.

[0029] The data acquisition module (10) is configured to collect clinical parameters from various data sources (such as information systems of clinics, pharmacies, health insurance, laboratories or other relevant data sources). The collected clinical parameters can be in the form of structured or unstructured data. The structured or unstructured data can include numerical data, text data, image data, audio data, video data, sensor data, etc. The clinical parameters include maternal parameters and fetal parameters. Maternal parameters include various medical and non-medical information, such as the mother's demographics, medical history, clinical history, health risk factors, etc. For example, maternal parameters can include the mother's age, race, height, weight, record of hypertension, smoking, diabetes, etc. Fetal parameters are preferably collected from ultrasound measurements obtained during the second trimester of pregnancy. For example, fetal parameters include gestational age, biparietal diameter, abdominal circumference, head circumference, femur length, amniotic fluid index, transverse cerebellar diameter, humeral length, cisterna magna size, estimated fetal weight, bronchial fold thickness, etc. Clinical parameters may be input into the data acquisition module (10) via a user interface such as a web interface or a mobile interface.

[0030] Figure 1b and Figure 1c An example of a user interface for inputting maternal and fetal parameters into a data acquisition module (10) is shown. Maternal parameters include data about the patient's demographics and medical history, such as race, smoking history, and others, while fetal parameters include data about fetal measurements obtained during an ultrasound scan, such as head circumference, amniotic fluid index, and the like.

[0031] The data encoding module (20) is connected to the data acquisition module (10). The data encoding module (20) is configured to execute a data encoding algorithm to transform the data of clinical parameters into different dimensions while retaining necessary medical information.

[0032] The processing module (30) is connected to the data encoding module (20). The processing module 30 is configured to perform data preprocessing on the collected clinical parameters. The processing module (30) extracts the clinical parameters and transforms them into a suitable data format to prepare the clinical parameters as input to the prediction model. The processing module (30) can perform various techniques or operations to preprocess the data. These techniques may include data interpolation, data standardization, data conversion, data cleaning, data normalization or scaling, feature selection, data transformation, data enhancement, etc.

[0033] The inference module (40) is connected to the processing module (30). The inference module (40) is configured to analyze clinical parameters using a predictive model to predict the risk of SGA condition in the fetus. The inference module (40) quantifies the risk as a probability score, wherein the probability score indicates the likelihood of the fetus being born with SGA condition or moderate gestational age or AGA condition. The inference module (40) may employ a suitable machine learning or deep learning algorithm as a predictive model. The machine learning or deep learning algorithm may utilize various architectures, techniques, and methods to learn patterns and make predictions based on data of clinical parameters. For example, the machine learning algorithm may include various algorithms such as linear regression, logistic regression, decision trees, support vector machines, k-nearest neighbors, random forests, etc., while the deep learning algorithm may include various algorithms utilizing artificial neural networks. In a preferred embodiment, the inference module (40) employs an eXtreme gradient boosting or XGBoost algorithm to generate the probability score. The XGBoost algorithm is an ensemble learning algorithm that combines the predictions of multiple weak decision tree models to create a stronger and more accurate final prediction.

[0034] To analyze the clinical parameters, the inference module (40) assigns the collected clinical parameters as input features. The inference module (40) calculates an importance score for each of the input features. Based on the importance score, the inference module (40) can further determine the causal factors for the prediction of a fetus with an SGA condition. The input feature with the highest importance score can indicate that the input feature contributes the most to the prediction of a fetus with an SGA condition. For example, assume that the inference module (40) predicts the fetus to be SGA, where the feature "hypertension" achieves the highest importance score. In this scenario, the inference module (40) can determine that the prediction of the fetus's SGA stems from the presence of maternal hypertension.

[0035] The data decoding module (50) is connected to the inference module (40). The data decoding module (50) is configured to execute a data decoding algorithm to transform the encoded clinical parameter data back to its original dimension.

[0036] The management module (60) is connected to the data decoding module (50). The management module (60) is configured to generate a diagnostic report and an intervention plan, wherein the intervention plan enables personalized recommendations or actions to be taken for monitoring and managing SGA or AGA fetuses. The intervention plan is generated based on the probability score and the diagnostic report associated with the predicted outcome. The intervention plan preferably complies with standard clinical guidelines, such as the Royal College of Obstetricians and Gynaecologists or RCOG guidelines for managing fetuses with SGA conditions, or the National Institute for Health and Care Excellence or NICE guidelines for managing antenatal care for fetuses with AGA conditions.

[0037] The display module (70) is connected to the management module (60). The display module (70) is configured to display data related to the prediction results, diagnostic reports and intervention plans to a user of the system (100). The display module (70) may include any device capable of displaying data via at least one user interface. For example, the display module (70) may be a personal computer, a tablet computer, a smart phone, a screen monitor, or other device capable of displaying a web interface or a mobile interface to present data. The display module (70) includes a set of interactive dashboards for providing graphical illustrations or indicators of the data, wherein the graphical illustrations or indicators may be in the form of charts, statistics, test reports, etc.

[0038] Figure 1d and Figure 1e An example of an interactive dashboard that displays prediction results, diagnostic reports, and intervention plans to a user of the system (100) via a display module (70) is shown. Figure 1d , the dashboard displays a probability score indicating the likelihood of a fetus being born with SGA status and a bar chart indicating the input features and importance scores used to determine the causal factors for the prediction of a fetus with SGA status. Figure 1e In the dashboard, a diagnostic report related to the predicted results and a recommended intervention plan for managing the predicted SGA fetus generated by the management module (60). The intervention plan includes recommended actions for the patient, such as scheduling frequent ultrasound scans to measure the condition of the fetus. Based on the results of the ultrasound scan, the medical staff can determine further actions to be taken. For example, if the results of the ultrasound scan show that the fetus is not growing well, the medical staff may make a plan for early delivery of the fetus.

[0039] Reference now Figure 2a, which shows a flow chart of a method for predicting and managing a SGA fetus according to an embodiment of the present invention. Initially, as in step 200, the data acquisition module (10) collects clinical parameters from various data sources, wherein the clinical parameters include maternal parameters and fetal parameters. The clinical parameters can be collected and input into the data acquisition module (10) via a user interface of a web application or a mobile application. Once the clinical parameters have been collected and input, the data acquisition module (10) sends the clinical parameters to the data encoding module (20).

[0040] Then, as in step 201, the data encoding module (20) performs a data encoding process on the clinical parameters using a data encoding algorithm. Once the data has been encoded, the data encoding module (20) sends the clinical parameters to the processing module (30). Then, as in step 202, the processing module (30) performs data preprocessing on the clinical parameters, wherein the data preprocessing is performed to extract the clinical parameters and transform them into a suitable data format for input to the prediction model. The processing module (30) can perform a variety of techniques or operations to preprocess the clinical parameters. These techniques can include data interpolation, data standardization, data conversion, data cleaning, data normalization or scaling, feature selection, data transformation, data enhancement, etc. Once the clinical parameters have been preprocessed, the processing module (30) sends the clinical parameters to the inference module (40).

[0041] The inference module (40) assigns the clinical parameters as input features and inputs the features into the prediction model. Then, as in step 203, the inference module (40) analyzes the clinical parameters to predict the risk of SGA or AGA condition in the fetus. The inference module (40) quantifies and generates the risk as a probability score, wherein the probability score indicates the likelihood of the fetus being born with an SGA condition or an AGA condition. In a preferred embodiment, the inference module (40) employs an XGBoost algorithm to generate the probability score. The XGBoost algorithm combines the predictions of multiple weak decision tree models to create a stronger and more accurate final prediction. In an alternative embodiment, whether the fetus will be born with an SGA or AGA condition is determined by comparing the probability score with a predetermined threshold. For example, if the probability score is higher than 0.5 or 50%, it indicates that the fetus is likely to be born with an SGA condition. On the other hand, if the probability score is lower than 0.5 or 50%, it indicates that the fetus is likely to be born with an AGA condition.

[0042] The inference module (40) also calculates an importance score for each of the input features. Based on the importance scores, the inference module (40) can further determine causal factors for the prediction of a fetus with an SGA condition. The input feature with the highest importance score can indicate that the input feature contributes most to the prediction of a fetus with an SGA condition. For example, assume that the inference module (40) predicts a fetus as SGA, where the input feature "hypertension" achieves the highest importance score. In this scenario, the inference module (40) can determine that the prediction of SGA for the fetus stems from the presence of maternal hypertension.

[0043] Thereupon, the inference module (40) sends the clinical parameters, importance scores, and probability scores to the data decoding module (50). The data decoding module (50) then performs a data decoding process using a data decoding algorithm as in step 204, wherein the data decoding algorithm transforms the encoded clinical parameter data back to its original dimensions. The data decoding module (50) also sends the clinical parameters, importance scores, and probability scores to the management module (60).

[0044] Then, as in step 205, the management module (60) analyzes the clinical parameters, importance scores, and probability scores to generate a diagnostic report and an appropriate intervention plan related to the predicted outcome. The intervention plan enables the management module (60) to generate recommendations for personalized prenatal care to monitor and manage SGA or AGA fetuses. The intervention plan preferably complies with standard clinical guidelines, such as the RCOG guidelines for the management of fetuses with SGA conditions and the NICE guidelines for the management of fetuses with AGA conditions. Figure 2b Describe in detail the sub-steps used to generate the intervention plan.

[0045] All data related to the predicted results, diagnostic reports, and intervention plans are displayed to the user of the system (100) via the user interface on the display module (70) as in step 206. The display module (70) includes a set of interactive dashboards for providing graphical illustrations or indicators of the data, where the graphical illustrations or indicators can be in the form of charts, statistical values, test reports, etc. The data displayed on the user interface can be used to provide the user of the system (100) with further insight into additional analysis, determination, or decision-making processes related to the management of the SGA or AGA fetus.

[0046] Reference now Figure 2b , which shows the use of Figure 2aThe flowchart of the sub-steps of generating an intervention plan made in step 205 of the method of FIG. Initially, the management module (60) obtains and analyzes the probability score as in step 2051. If the probability score indicates that the fetus is more likely to be born with an AGA condition, the management module (60) outputs a recommendation on a course of action to be taken to care for the AGA fetus in accordance with the NICE guidelines as in step 2052.

[0047] On the other hand, if the probability score indicates that the fetus is more likely to be born with an SGA condition, the management module (60) obtains the collected clinical parameters of the fetus. Thereafter, as in step 2053, the management module (60) analyzes the fetal biometric measurements. The fetal biometric measurements provide data on anatomical parameters of the fetus, such as biparietal diameter, abdominal circumference, head circumference, femur length, etc. Next, as in step 2054, the management module (60) further analyzes the fetal Doppler measurements. The Doppler measurements include umbilical artery or UA Doppler, middle cerebral artery or MCA Doppler, and ductus venosus or DV Doppler. The Doppler measurements provide data on blood flow patterns and velocities within various blood vessels of the fetus, such as the resistance index or RI, the pulsatility index or PI, blood flow velocity within the blood vessels, etc.

[0048] During the analysis, data regarding the fetal biometric measurements and Doppler measurements are compared to predefined thresholds. Based on the results of the analysis, the management module (60) then generates an ultrasound recommendation, as in step 2055. In an example embodiment, the ultrasound recommendation includes a timing schedule for performing ultrasound measurements and Doppler measurements. Next, the management module (60) generates a delivery recommendation, as in step 2056. In an example embodiment, the delivery recommendation includes a timing schedule, the administration of steroids, and a method of fetal delivery.

[0049] Figure 2cAn example of an intervention plan for managing an SGA fetus generated by a management module (60) is shown. During analysis of fetal biometric measurements, the management module (60) determines that the value of abdominal circumference or estimated fetal weight is less than the 10th custom percentile, or that the continuous measurement indicates fetal growth restriction or FGR. The management module (60) then proceeds to analyze the UA Doppler measurements of the fetus. Based on the fetal biometric and Doppler measurements, the management module (60) then generates appropriate ultrasound recommendations and delivery recommendations. For example, if the UA Doppler measurement records that absent / reversed end-diastolic velocities or AREDV are observed, the management module (60) outputs a recommendation for weekly ultrasound scans to measure abdominal circumference or AC and estimated fetal weight or EFW. The management module (60) also outputs a recommendation for daily UA Doppler and DV Doppler measurements. The management module (60) then proceeds to output recommendations regarding a delivery schedule, such as recommending delivery before 32 weeks with the administration of steroids.

[0050] Although not elaborated herein, it should be understood that the methods of the present invention may include additional necessary steps for predicting and managing an SGA fetus. For example, prior to implementing a machine learning or deep learning algorithm for SGA prediction, the inference module (40) may be involved in training, testing, and evaluating the algorithm. Further to this, while the process is described as involving automatic generation by the management module (60) of recommendations for medical personnel to manage an SGA or AGA fetus, it should be recognized that additional insights or manual input from a fetal medicine expert may be integrated to help improve or enhance the intervention plan for an SGA or AGA fetus.

[0051] Although the embodiments of the present invention have been shown and described, these embodiments are not intended to illustrate and describe all possible forms of the present invention. Instead, the words used in the specification are words of description rather than limitation, and various changes can be made without departing from the scope of the present invention.

Claims

1. A system (100) for predicting and managing a small for gestational age (SGA) or SGA fetus, comprising: a) a data acquisition module (10) configured to collect a plurality of clinical parameters; b) a processing module (30) configured to perform data preprocessing on the plurality of clinical parameters, Characterized in that the system (100) further comprises: c) an inference module (40) configured to analyze the plurality of clinical parameters using a prediction model to predict the risk of SGA condition in the fetus, wherein the inference module (40) is connected to the processing module (30).

2. The system (100) according to claim 1, wherein: The system (100) further comprises a management module (60) configured to generate a diagnostic report and at least one intervention plan for managing and monitoring the SGA or average-for-gestational-age (AGA) or AGA fetus.

3. The system (100) of claim 1, wherein: The system (100) further comprises a display module (70) configured to display data related to the prediction result, the diagnosis report and the at least one intervention plan to a user of the system (100).

4. The system (100) of claim 1, wherein: The system (100) also includes a data encoding module (20) configured to execute a data encoding algorithm.

5. The system (100) of claim 1, wherein: The system (100) also includes a data decoding module (50) configured to execute a data decoding algorithm.

6. The system (100) of claim 1, wherein: The inference module (50) quantifies the risk into a probability score and calculates importance scores for a plurality of input features.

7. A method for predicting and managing a small-for-gestational-age (SGA) or SGA fetus, It is characterized by the following steps: a) collecting a plurality of clinical parameters through a data acquisition module (10); b) analyzing the plurality of clinical parameters by an inference module (40) to predict the risk of SGA or AGA for gestational age or AGA condition in the fetus; as well as c) generating, by a management module (60), a diagnostic report and at least one intervention plan associated with the prediction of SGA or AGA condition in said fetus.

8. The method according to claim 7, wherein: The method further comprises: performing a step of data encoding the plurality of clinical parameters by a data encoding module (20).

9. The method according to claim 7, wherein: The method further comprises: performing data preprocessing on the plurality of clinical parameters by a processing module (30).

10. The method according to claim 7, wherein: The method further comprises: performing a step of data decoding on the plurality of clinical parameters by a data decoding module (50).

11. The method according to claim 7, wherein: The method further comprises the step of displaying data related to the prediction result, the diagnosis report and the at least one intervention plan to a user of the system (100) via a display module (70).

12. The method according to claim 7, wherein: The step of analyzing the plurality of clinical parameters by the inference module (40) to predict the risk of SGA or AGA condition in the fetus further comprises the following steps: a) assigning the plurality of clinical parameters as input features; b) inputting the input features into a prediction model; c) quantifying the risk as a probability score; and d) Calculating importance scores for the input features.

13. The method according to claim 7, wherein: The step of generating the at least one intervention plan by the management module (60) further comprises the following steps: a) Analyze the probability scores; b) analyzing fetal biometric measurements if the probability score indicates that the fetus is more likely to be born with the SGA condition; c) analyzing fetal Doppler measurements; d) generating ultrasound recommendations; and e) Generate delivery recommendations.

14. The method according to claim 13, wherein: If the probability score indicates that the fetus is more likely to be born with the AGA condition, a recommendation is output for a course of action to be taken for caring for the AGA fetus.

Images