An artificial intelligence-based risk detection system using fetal sound signals acquired from an NST (non-stress test) device

Abstract

The present invention relates to an artificial intelligence-based risk detection system using fetal sound signals acquired from an NST device, wherein said system is used in the medical field to analyze fetal heart rates in a pregnant subject via an NST (Non-Stress Test) device and to monitor whether the life course of the fetus and the mother is proceeding normally, thereby evaluating, diagnosing, and generating alerts regarding the presence of any fetal distress affecting the maternal health status or the growth and development of the fetus.

Description

[0001] AN ARTIFICIAL INTELLIGENCE-BASED RISK DETECTION SYSTEM USING FETAL SOUND SIGNALS ACQUIRED FROM AN NST (NON-STRESS TEST) DEVICE

[0002] Technical Field

[0003] The present invention relates to an artificial intelligence-based risk detection system using fetal sound signals acquired from an NST device, wherein said system is used in the medical field to analyze fetal heart rates in a pregnant subject via an NST (Non-Stress Test) device and to monitor whether the life course of the fetus and the mother is proceeding normally, thereby evaluating, diagnosing, and generating alerts regarding the presence of any fetal distress affecting the maternal health status or the growth and development of the fetus.

[0004] Background of the Invention

[0005] Ensuring a healthy development of the fetus within the womb is crucial, as any abnormal condition could pose a serious risk to both the fetus and the mother. For this reason, regular Non-Stress Tests (NSTs) are vital. An NST may be defined as an observational test conducted during pregnancy to assess whether the fetal heart rate, responses, and movements are within healthy parameters. Since it doesn’t induce stress on the fetus, it is referred to as a non-stress test. In certain pregnancy conditions, NST becomes even more critical compared to normal circumstances. These conditions include:

[0006] • If an abnormality is detected in the fetus’s growth and development, NST monitoring helps determine if the fetus returns to a normal course of development;

[0007] • If the mother is diagnosed with conditions like gestational diabetes, preeclampsia, or hypertension, NST monitoring tracks the healthy progression of both mother and fetus;

[0008] • If the fetus shows a slow heart rate or delayed responses, close monitoring through NST is required;

[0009] • In cases of prior miscarriage, multiple gestation, or advanced maternal age, monitoring both the mother and fetus closely is essential;

[0010] • Additionally, if the fetus is diagnosed with a disease or there is blood incompatibility between the mother and fetus, monitoring is needed due to the increased risk. Medical specialists generally recommend initiating NST monitoring from the 32nd week of pregnancy onward. Prior to the 32nd week, the neural connections between the fetal nervous system and the heart may not have fully matured. As a result, the fetal response to movement is typically interpreted through changes in the fetal heart rate.

[0011] Within an NST device, desired parameters are monitored in detail on a display commonly referred to as a fetal monitor. Based on these parameters, various data such as the fetal heart rate count, heart rate duration, and the number of uterine contractions of the mother can be observed. NST devices typically employ two probes, namely a TOCO probe and a US probe. The TOCO probe detects the number and intensity of uterine contractions, while the US probe monitors the rate and duration of fetal heartbeats for transmission to the fetal monitor.

[0012] The NST TOCO value may be understood as a parameter that tracks the number, duration, and intensity of uterine contractions. Although this examination is generally monitored close to the time of delivery, in certain cases, particularly in high-risk pregnancies, it may be initiated earlier. This is because, in situations involving risks such as preterm labor, monitoring uterine contractions plays a critical role in preventing fetal hypoxia. Through NST, diagnoses of fetal distress (including tachycardia, bradycardia, late deceleration, and variable deceleration) may be established based on fetal heart sounds. Under such conditions, the sound levels and rhythms of fetal heartbeats obtained from the NST device exhibit changes. Definitions of fetal distress conditions are provided below:

[0013] Tachycardia: A condition in which the fetal heart rate is 160 beats per minute or higher over a ten-minute period. Moderate tachycardia ranges from 160 to 180 beats per minute, whereas severe tachycardia exceeds 180 beats per minute.

[0014] Bradycardia :A condition in which the fetal heart rate falls below 120 beats per minute over a ten-minute period. Mild bradycardia ranges between 100 and 120 beats per minute, moderate bradycardia is below 100 beats per minute, and severe bradycardia is below 70 beats per minute.

[0015] Early deceleration: A deceleration that occurs synchronously with uterine contractions, wherein the deepest point of the deceleration coincides with the peak of the contraction. Early deceleration is generally physiological and most commonly occurs when the fetal head is pressed against the perineum. Its occurrence at the onset of labor may indicate cephalopelvic disproportion. In early deceleration, rhythm disturbances are typically of shorter duration. In the literature, decreases in fetal heart rate associated with early deceleration exhibit onset-to-peak durations of 30 seconds or longer.

[0016] Late Deceleration: The fetal heart rate decreases when the blood flow from the mother to the placenta is compromised. This deceleration does not occur immediately upon the onset of uterine contractions. As the pressure from the uterine contractions increases, the fetal heart rate continues to drop, and the fetus’s oxygen levels decrease. In late deceleration, the symmetrical decrease and recovery of the heart rate extend beyond the onset and peak durations of 30 seconds.

[0017] Variable Deceleration: Variable deceleration occurs without a temporal or morphological relationship between uterine contractions and the fetal heart rate. This deceleration is caused by umbilical cord compression, which leads to fetal hypoxia. The variability in the deceleration is due to the varying degrees of umbilical cord compression during each contraction. The sudden drops in the fetal heart rate are less than 30 seconds for both onset and peak. The fetal heart rate decrease may range between 15 seconds to 2 minutes.

[0018] Midwives are responsible for monitoring the health of both the mother and the fetus throughout labor. When complications or emergencies arise, timely diagnosis and appropriate intervention by the healthcare team are required. The ability of midwives to accurately identify risks and promptly communicate them to attending physicians plays a critical role in preventing obstetric errors.

[0019] In the prior art, all evaluations and diagnoses are performed by healthcare professionals such as midwives and physicians. However, assessments based solely on human judgment are susceptible to error, which may place both the mother and the fetus at risk while also imposing a substantial level of responsibility on healthcare personnel. In addition, inefficient workforce planning can result in the loss of human resources.

[0020] An examination of studies in the prior art indicates that, due to the limitations of existing NST techniques, there is a need for improvements in this field with respect to maternal and fetal health and comfort, the psychological well-being and comfort of healthcare personnel, and human resources management. Description of the Invention

[0021] In the following description, the artificial intelligence-based risk detection system using fetal sound signals acquired from an NST (Non-Stress Test) device is provided solely for facilitating a clearer understanding of the subject matter, without any intention to impose a limiting effect.

[0022] The invention relates to an artificial intelligence-based risk detection system using fetal sound signals acquired from an NST device, wherein said system is used in the medical field to analyze fetal heart rates in a pregnant subject via an NST (Non-Stress Test) device and to monitor whether the life course of the fetus and the mother is proceeding normally, thereby evaluating, diagnosing, and generating alerts regarding the presence of any fetal distress affecting the maternal health status or the growth and development of the fetus.

[0023] Preferably, the invention provides an artificial intelligence-based risk detection system that uses fetal heart rates acquired from the NST (Non-Stress Test) device to evaluate the presence of any fetal distress affecting the maternal health status or the growth and development of the fetus.

[0024] More preferably, the invention provides an artificial intelligence-based risk detection system that uses signals acquired from the NST device to evaluate, diagnose, and generate alerts regarding the presence of any fetal distress affecting the maternal health status or the growth and development of the fetus.

[0025] Even more preferably, the invention provides an artificial intelligence-based risk detection system using the fetal sound signals acquired from the NST (Non-Stress Test) device.

[0026] The primary object of the invention is to enable the diagnosis of fetal distress within 15-30 seconds using artificial intelligence technologies without reliance on human intervention.

[0027] Another object of the invention is to eliminate the error potential inherent in human-based evaluations, thereby reducing risks to both the mother and the fetus.

[0028] A further object of the invention is to remove the high level of responsibility that is typically imposed on healthcare workers. Still a further object of the invention is to provide a solution-oriented system that facilitates effective human resource planning and contributes to achieving successful outcomes.

[0029] Figure 1 illustrates the operation principle and workflow of the artificial intelligence-based risk detection system using fetal sound signals acquired from the NST device. The process begins when the pregnant subject is connected to the NST device, followed by the activation of the start module. Fetal heartbeats are recorded on a computer system. The recorded data is classified based on characteristics such as noise, magnitude, format, intensity, duration, and rhythm to create a dataset. The dataset is then divided into 80% training data and 20% test data, wherein the trained system then uses machine learning algorithms, including Naive Bayes (NB), Random Forest (RF), Decision Tree (DT), k-Nearest Neighbors (k-NN), and Support Vector Machines (SVM), so as to classify results based on fetal distress, thereby diagnosing tachycardia, bradycardia, late deceleration, early deceleration, or variable deceleration. The system provides a fetal distress alert retroactively within 31 seconds after a 30-second decrease in fetal heart rate. The system also diagnoses fetal distress when the fetal heart rate exceeds 160 beats per minute or falls below 120 beats per minute. Following the diagnosis, a message alert module notifies the authorized healthcare personnel via SMS and e-mail and registers the diagnosis within the patient and hospital information system. The system also includes an audible alert system. The diagnoses are automatically recorded in the electronic archive within the hospital information system. Accordingly, the present invention provides decision support for experts and enables the early detection of potential health conditions.

Claims

CLAIMS1. A system for analyzing fetal heart rates in a pregnant subject via an NST (NonStress Test) device and monitoring whether the life course of the fetus and the mother is proceeding normally, thereby evaluating, diagnosing, and generating alerts regarding the presence of fetal distress, characterized in that said system applies machine-learning algorithms, including Naive Bayes (NB), Random Forest (RF), Decision Tree (DT), k-Nearest Neighbors (k-NN), and Support Vector Machines (SVM), so as to diagnose fetal distress and generate an alert based on fetal sound signals acquired from the NST device.

2. A system according to claim 1 , characterized in that the system generates a dataset by classifying the fetal sound signals acquired from the NST device according to noise, magnitude, format, waveform, intensity, duration, and rhythm thereof, and uses this dataset in the machine-learning algorithms by allocating 80% as training data and 20% as test data.

3. A system according to claim 1 or claim 2, characterized in that the system applies machine-learning algorithms, including Naive Bayes (NB), Random Forest (RF), Decision Tree (DT), k-Nearest Neighbors (k-NN), and Support Vector Machines (SVM), so as to diagnose fetal distress by classifying it as tachycardia, bradycardia, late deceleration, early deceleration, or variable deceleration.

4. A system according to claim 1 or claim 3, characterized in that the system transmits the fetal distress diagnosis to authorized healthcare personnel and to a Hospital Information System via SMS and e-mail, and generates alerts through an alert module.

Images