Federated Machine Learning Based Fetal Health Prediction Empowered with Bio-Signal Cardiotocography

Cardiotocography measures the fetal heart rate in the fetus during pregnancy to ensure physical health because cardiotocography gives data about fetal heart rate and uterine shrinkages which is very beneficial to detect whether the fetus is normal or suspect or pathologic.Various cardiotocography measures infer wrongly and give wrong predictions because of human error.The traditional way of reading the cardiotocography measures is the time taken and belongs to numerous human errors as well.Fetal condition is very important to measure at numerous stages and give proper medications to the fetus for its well-being.In the current period Machine learning(ML)is a well-known classification strategy used in the biomedical field on various issues because ML is very fast and gives appropriate results that are better than traditional results.ML techniques play a pivotal role in detecting fetal disease in its early stages.This research article uses Federated machine learning(FML)and ML techniques to classify the condition of the fetus.This study proposed a model for the detection of bio-signal cardiotocography that uses FML and ML techniques to train and test the data.So,the proposed model of FML used numerous data preprocessing techniques to overcome data deficiency and achieves 99.06%and 0.94%of prediction accuracy and misprediction rate,respectively,and parallel the proposed model applying K-nearest neighbor(KNN)and achieves 82.93%and 17.07%of prediction accuracy and misprediction accuracy,respectively.So,by comparing both models FML outperformed the KNN technique and achieved the best and most appropriate prediction results as compared with previous studies the proposed study achieves the best and most accurate results.

Dynamic Multi-Layer Perceptron for Fetal Health Classification Using Cardiotocography Data

Fetal health care is vital in ensuring the health of pregnant women and the fetus.Regular check-ups need to be taken by the mother to determine the status of the fetus’growth and identify any potential problems.To know the status of the fetus,doctors monitor blood reports,Ultrasounds,cardiotocography(CTG)data,etc.Still,in this research,we have considered CTG data,which provides information on heart rate and uterine contractions during pregnancy.Several researchers have proposed various methods for classifying the status of fetus growth.Manual processing of CTG data is time-consuming and unreliable.So,automated tools should be used to classify fetal health.This study proposes a novel neural network-based architecture,the Dynamic Multi-Layer Perceptron model,evaluated from a single layer to several layers to classify fetal health.Various strategies were applied,including pre-processing data using techniques like Balancing,Scaling,Normalization hyperparameter tuning,batch normalization,early stopping,etc.,to enhance the model’s performance.A comparative analysis of the proposed method is done against the traditional machine learning models to showcase its accuracy(97%).An ablation study without any pre-processing techniques is also illustrated.This study easily provides valuable interpretations for healthcare professionals in the decision-making process.

Machine Learning Prediction of Fetal Health Status from Cardiotocography Examination in Developing Healthcare Contexts

Reducing neonatal mortality is a critical global health objective,especially in resource-constrained developing countries.This study employs machine learning(ML)techniques to predict fetal health status based on cardiotocography(CTG)examination findings,utilizing a dataset from the Kaggle repository due to the limited comprehensive healthcare data available in developing nations.Features such as baseline fetal heart rate,uterine contractions,and waveform characteristics were extracted using the RFE wrapper feature engineering technique and scaled with a standard scaler.Six ML models—Logistic Regression(LR),Decision Tree(DT),Random Forest(RF),Gradient Boosting(GB),Categorical Boosting(CB),and Extended Gradient Boosting(XGB)—are trained via cross-validation and evaluated using performance metrics.The developed models were trained via cross-validation and evaluated using ML performance metrics.Eight out of the 21 features selected by GB returned their maximum Matthews Correlation Coefficient(MCC)score of 0.6255,while CB,with 20 of the 21 features,returned the maximum and highest MCC score of 0.6321.The study demonstrated the ability of ML models to predict fetal health conditions from CTG exam results,facilitating early identification of high-risk pregnancies and enabling prompt treatment to prevent severe neonatal outcomes.

Admission Cardiotocography: Its Role in Predicting Perinatal Outcome in Term, Uncomplicated (Low Risk) Pregnant Women in Spontaneous Labour

Introduction: Labour admission cardiotocography (CTG) is commonly used non-invasive method of fetal monitoring in Sri Lanka. It may have a potentialto predict perinatal outcome in low-risk term pregnancies. Objectives: Objectives of the study were to determine the perinatal outcomes of normal, suspicious and pathological admission CTGs and role of labour admission cardiotocography as a predictive test for perinatal outcome in low-risk term pregnancies in spontaneous labour. Methods: This study was a prospective observational study done involving 445 low risk, term pregnancies in spontaneous labour. Labour admission CTG was performed in each pregnancy and categorized into normal, suspicious and pathological CTG according to criteria depicted by National Institute of Clinical Excellence (NICE) guideline 2007. Apgar score less than 7 at five minutes, resuscitation at birth, admission to neonatal intensive care unit (NICU), seizure within first 24 hours of birth and meconium-stained amniotic fluid were the primary outcome measures to assess fetal asphyxia. Mode of delivery in each category, nuchal cord at birth were also assessed. Results: Majority of participants were in 25-to-29-year age group and were nulliparous. Frequencies of normal, suspicious and pathological CTG were 74.8%, 18% and 7.2% respectively. Pathological CTG was significantly associated with low Apgar score compared to non-pathological CTG group (p 0.005) while other outcome measures were not significant. Rate of operative delivery was 68% in pathological group and 20.8% in non-pathological CTG group. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of labour admission CTG to detect fetal asphyxia were 51.85%, 95.69%, 43.75% and 96.85% respectively. Conclusions: Incidence of pathological labour admission CTG was 7.2%. Apgar score less than 7 at five minutes of birth was significantly associated with pathological CTG group compared to non-pathological CTG (p 0.05). Worsening of CTG from normal to pathological showed increasing rate of operative delivery. Even though sensitivity and positive predictive values of labour admission CTG were low, specificity and negative predictive values were high for detecting low Apgar score. Therefore, labour admission CTG has a value in excluding adverse perinatal outcomes in low-risk term pregnancies in spontaneous labour.

Increasing rate of Caesarean Section Due to Non-Reassuring Cardiotocography

Objective: To evaluate increasing rate of caesarean section due to non-reassuring cardiotocography. Methods: This study is carried out in obs/gyn department of Liaquat university hospital from 2012 to 2013. After permission from ERC, patients enrolled for study meeting inclusion criteria with non-reactive cardiotocography undergo caesarean section, and results are analysis through SSPS version 17. Results: There was wide variation of maternal age ranging from a minimum of 20 years to 30 years. The mean age was 26 ± 2.1 years. In our study mostly patients were primigravida 58 (58%) between 2 - 4 were 22 (22%) more than para 5 were 20 (20%) patients. In our study mostly patients undergone caesarean section 81 (81%) 19 delivered vaginally (19%). In our study the gestational age was >37 weeks, ranging from a minimum of 37 weeks to 42 weeks. The mean age was 37 + 2.4 week. Mostly patients observed 37 - 38 wks in (52.67%), 39 - 40 wks in (32.14%) and 41 - 42 wks in (15.17%). In our study mostly Apgar score were more than 7 was 63 (63%) cases and less than 7 Apgar score in 37 (37%). Conclusion: Cardiotocography is a useful and indispensable adjunct to monitor the condition of endangered fetus. However, there is a need to develop a standardized and unambiguous definition of FHR tracing to reduce the incidence of false positive findings that may result in increased incidence of unnecessary intervention particularly caesarean section.

Computerized Analysis of Antepartum Cardiotocography: A Review

Cardiotocography measures the human fetal heart rate and uterine activity using ultrasound.While it has been a mainstay in antepartum care since the 1960s,cardiotocograms consist of complex signals that have proven difficult for clinicians to interpret accurately and as such clinical inference is often difficult and unreliable.Previous attempts at codifying approaches to analyzing the features within these signals have failed to demonstrate reliability or gain sufficient traction.Since the early 1990s,the Dawes-Redman system of automated computer analysis of cardiotocography signals has enabled robust analysis of cardiotocographic signal features,employing empirically-derived criteria for assessing fetal wellbeing in the antepartum.Over the past 30 years,the Dawes-Redman system has been iteratively updated,now incorporating analyses from over 100,000 pregnancies.In this review,we examine the history of cardiotocography,signal processing methodologies and feature identification,the development of the Dawes-Redman system,and its clinical applications.

Interobserver Agreement and Reliability of Intrapartum Nonreassuring Cardiotocography and Prediction of Neonatal Acidemia

Objective:To evaluate the agreement and reliability of intrapartum nonreasurring cardiotocography(CTG)interpretation and prediction of neonatal acidemia by obstetricians working in different centers.Methods:A retrospective cohort study involving two tertiary hospitals(The First Affiliated Hospital of Sun Yat-sen University and Perking University Third Hospital)was conducted between 30th September 2018 and 1st April 2019.Six obstetricians from two hospitals with three levels of experience(junior,medium,and senior)reviewed 100 nonreassuring fetal heart rate(FHR)tracings from 1 hour before the onset of abnormalities until delivery.Each reviewer determined the FHR pattern,the baseline,variability,and presence of acceleration,deceleration,sinusoidal pattern,and predicted whether neonatal acidemia and abnormal umbilical arterial pH<7.1 would occur.Inter-observer agreement was assessed using the proportions of agreement(Pa)and the proportion of specific agreement(Pa for each category).Reliability was evaluated with the kappa statistic(k-Light’s kappa for n raters)and Gwet’s AC1 statistic.Results:Good inter-observer agreement was found in evaluation of most variables(Pa>0.5),with the exception of early deceleration(Pa=0.39,95%confidence interval(CI):0.36,0.43).Reliability was also good among most variables(AC1>0.40),except for acceleration,early deceleration,and prediction of neonatal acidemia(AC1=0.17,0.10,and 0.25,respectively).There were no statistically significant differences among the three groups,except in the identification of accelerations(Pa=0.89,95%CI:0.83,0.95;Pa=0.50,95%CI:0.41,0.60,and Pa=0.35,95%CI:0.25,0.43 in junior,medium and senior groups,respectively)and the prediction of neonatal acidemia(Pa=0.52,0.52,and 0.62 in junior,medium and senior groups,respectively),where agreement was highest and lowest in the junior-level group,respectively.The accuracy and sensitivity of the prediction for umbilical artery pH<7.1 were similar among the three groups,but the specificity was higher in the senior groups(93.68%vs.92.53%vs.98.85%in junior,medium and senior groups,P=0.015).Conclusion:Although we found a good inter-observer agreement in the evaluation of the most basic CTG features and FHR category statistically,it was insufficient to meet the clinical requirements for"no objection"interpretation for FHR tracings.Further specialized training is needed for standardized interpretation of intrapartum FHR tracings.

Ensemble Learning for Fetal Health Classification

Cardiotocography(CTG)represents the fetus’s health inside the womb during labor.However,assessment of its readings can be a highly subjective process depending on the expertise of the obstetrician.Digital signals from fetal monitors acquire parameters(i.e.,fetal heart rate,contractions,acceleration).Objective:This paper aims to classify the CTG readings containing imbalanced healthy,suspected,and pathological fetus readings.Method:We perform two sets of experiments.Firstly,we employ five classifiers:Random Forest(RF),Adaptive Boosting(AdaBoost),Categorical Boosting(CatBoost),Extreme Gradient Boosting(XGBoost),and Light Gradient Boosting Machine(LGBM)without over-sampling to classify CTG readings into three categories:healthy,suspected,and pathological.Secondly,we employ an ensemble of the above-described classifiers with the oversamplingmethod.We use a random over-sampling technique to balance CTG records to train the ensemble models.We use 3602 CTG readings to train the ensemble classifiers and 1201 records to evaluate them.The outcomes of these classifiers are then fed into the soft voting classifier to obtain the most accurate results.Results:Each classifier evaluates accuracy,Precision,Recall,F1-scores,and Area Under the Receiver Operating Curve(AUROC)values.Results reveal that the XGBoost,LGBM,and CatBoost classifiers yielded 99%accuracy.Conclusion:Using ensemble classifiers over a balanced CTG dataset improves the detection accuracy compared to the previous studies and our first experiment.A soft voting classifier then eliminates the weakness of one individual classifier to yield superior performance of the overall model.

Imperative for improvements and international convergence of intrapartum fetal monitoring: A bird's eye view

Intrapartum fetal monitoring has been criticized for the lack of evidence of improvement in fetal outcome despite causing increased operative intervention. Paradoxically, cardiotocography （CTG） has been a major driver for litigation for neonatal neurological injury. This analytical review tries to explore why extensive clinical studies and trials over 50 years have failed to demonstrate or bring about signifcant improvement in intrapartum fetal monitoring. There seems a need for significant reform. International congruence on most aspects of CTG interpretation [defnitions of fetal heart rate （FHR） parameters, CTG recording speed, 3-tier systems, etc .] is highly desirable to facilitate future meaningful clinical studies, evaluation and progress in this field. The FHR changes are non-specific and poor surrogate for fetal well-being. As a compromise for maintaining low false-negative results for fetal acidemia, a high false-positive value may have to be accepted. The need for redefning the place of adjuvant tests of fetal well-being like fetal blood sampling or fetal electrocardiography （ECG） is discussed. The FHR decelerations are often deterministic （center-stage） in CTG interpretation and 3-tier categorization. It is discussed if their scientifc and physiological classifcation （avoiding framing and confirmation biases） may be best based on time relationship to uterine contractions alone. This may provide a more sound foundation which could improve the reliability and further evolution of 3-tier systems. Results of several trials of fetal ECG （STAN） have been inconclusive and a need for a fresh approach or strategy is considered. It is hoped that the long anticipated Computer-aided analysis of CTG will be more objective and reliable （overcome human factors） and will offer valuable support or may eventually replace visual CTG interpretation. In any case, the recording and archiving all CTGs digitally and testing cord blood gases routinely in every delivery would be highly desirable for future research. This would facilitate well designed retrospective studies which can be very informative especially when prospective randomised controlled trials are often diffcult and resource-intensive.

Fetal distress prediction using discriminant analysis, decision tree, and artificial neural network

Fetal distress is one of the main factors to cesarean section in obstetrics and gynecology. If the fetus lack of oxygen in uterus, threat to the fetal health and fetal death could happen. Cardiotocography (CTG) is the most widely used technique to monitor the fetal health and fetal heart rate (FHR) is an important index to identify occurs of fetal distress. This study is to propose discriminant analysis (DA), decision tree (DT), and artificial neural network (ANN) to evaluate fetal distress. The results show that the accuracies of DA, DT and ANN are 82.1%, 86.36% and 97.78%, respectively.

Critical evaluation of American categorization of fetal heart rate (FHR) decelerations and three tier classification—Shortcomings, contradictions, remedies and need for debate

Fetal heart rate (FHR) decelerations are the commonest aberrant feature on cardiotocograph (CTG) thus having a major influence on classification ofFHRpatterns into the three tier system. The unexplained paradox of early decelerations (head compression—an invariable phenomenon in labor) being extremely rare [1] should prompt a debate about scientific validity of current categorization. This paper demonstrates that there appear to be major fallacies in the pathophysiological hypothesis (cord compression—baroreceptor mechanism) underpinning of vast majority of (variable?) decelerations. Rapid decelerations during contractions with nadir matching peak of contractions are consistent with “pure” vagal reflex (head compression) rather than result of fetal blood pressure or oxygenation changes from cord compression. Hence, many American authors have reported that the abrupt FHR decelerations attributed to cord compression are actually due to head compression [2-6]. The paper debates if there are major fundamental fallacies in current categorization of FHR decelerations based concomitantly on rate of descent (reflecting putative aetiology?) and time relationship to contractions. Decelerations with consistently early timing (constituting majority) seem to get classed as “variable” because of rapid descent. A distorted unscientific categorization of FHR decelerations could lead to clinically unhelpful three tier classification system. Hence, the current unphysiological classification needs a fresh debate with consideration of alternative models and re-evaluation of clinical studies to test these. Open debate improves patient care and safety. The clue to benign reflex versus hypoxic nature of decelerations seems to be in the timing rather than the rate of descent. Although the likelihood of fetal hypxemia is related to depth and duration ofFHRdecelerations, the cut-offs are likely to be different for early/late/variable decelerations and it seems to be of paramount importance to get this discrimination right for useful visual or computerized system of CTG interpretation.

Evolving Frameworks for the Foundation and Practice of Electronic Fetal Monitoring

Since the 1970s,electronic fetal monitoring(EFM)also known as cardiotocography(CTG)has been used extensively in labor around the world,despite its known failure to help prevent many babies from developing neonatal encephalopathy and cerebral palsy.Part of EFM’s poor performance with respect to these outcomes arises from a fundamental misunderstanding of the differences between screening and diagnostic tests,subjective classifications of fetal heart rate(FHR)patterns that lead to large inter-observer variability in its interpretation,failure to appreciate early stages of fetal compromise,and poor statistical modeling for its use as a screening test.We have developed a new approach to the practice and interpretation of EFM called the fetal reserve index(FRI)which does the following:(1)breaking FHR components down into 4 components,(heart rate,variability,accelerations,and decelerations);(2)contextualizing the metrics by adding increased uterine activity;(3)adding specific maternal,fetal,and obstetric risk factors.The result is an eight-point scoring metric that,when directly compared with current American College of Obstetricians and Gynecologists EFM categories,even in version 1.0 with equal weighting of variables,shows that the FRI has performed much better for identifying cases at risk before damage had occurred,reducing the need for emergency deliveries,and lowering overall Cesarean delivery rates.With increased data,we expect further refinements in the specifics of scoring that will allow even earlier detection of compromise in labor.