On usage of artificial intelligence for predicting mortality during and post-pregnancy: a systematic review of literature

Application of machine learning in identifying risk factors for low APGAR scores

BACKGROUND

Identifying the risk factors for low APGAR scores at birth is critical for improving neonatal outcomes and guiding clinical interventions.

METHODS

This study aimed to develop a machine-learning model that predicts low APGAR scores by incorporating maternal, fetal, and perinatal factors in Wad Medani, Sudan. Using a Random Forest Classifier, we performed hyper-parameter optimization through Grid Search cross-validation (CV) to identify the best-performing model configuration.

RESULTS

The optimized model achieved excellent predictive performance, as evidenced by high F1 scores, accuracy, and balanced precision-recall metrics on the test set. In addition to prediction, feature importance analysis was conducted to identify the most influential risk factors contributing to low APGAR scores. Key predictors included gestational age, maternal BMI, mode of delivery, and history of previous complications such as stillbirth or abortion. Using 5-fold cross-validation (CV), the random forest model performance scored accuracy at 96%, precision at 98%, recall at 97%, and F1-score at 97% when classifying infants with APGAR score.

CONCLUSION

This study underscores the importance of incorporating machine learning approaches in obstetric care to understand better and mitigate the risk factors associated with adverse neonatal outcomes, particularly low APGAR scores. The results provide a foundation for developing targeted interventions and improving prenatal care practices.

Artificial Intelligence for Women and Child Healthcare: Is AI Able to Change the Beginning of a New Story? A Perspective

Background and Aims

Maternal and neonatal mortality remain critical global health challenges, particularly in low-resource settings where preventable deaths occur due to inadequate access to timely care. This article explores the potential of Artificial Intelligence (AI) to enhance maternal and child healthcare by improving early risk identification, diagnosis, treatment recommendations, and postpartum monitoring.

Methods

It explores the use of AI in identifying pregnancy-related risks, recommending treatments, predicting adverse outcomes, and monitoring postpartum and neonatal care. Various AI models, including supervised machine learning, Large Language Models (LLMs), and Small/Medium Language Models (SLMs/MLMs), are discussed in terms of their feasibility into resource-limited healthcare systems.

Results

AI has demonstrated significant potential in identifying pregnancy-related risks, recommending treatments, predicting adverse outcomes, and supporting postpartum and neonatal care. While AI-driven solutions can optimize healthcare decision-making and resource allocation, challenges such as data availability, integration into clinical workflows, and ethical considerations must be addressed for widespread adoption.

Conclusion

AI offers promising solutions to reduce maternal and neonatal mortality by enhancing risk detection and clinical decision-making. However, its real-world implementation requires overcoming barriers related to data quality, infrastructure, and equitable deployment. Future efforts should focus on data standardization, AI model optimization for resource-limited settings, and ethical considerations in clinical integration.

Patients' knowledge, attitudes, and practices regarding lifestyle related dry eye

Rising incidences of dry eye, often attributed to modern lifestyle and environmental factors, highlight the need for knowledge, attitudes, and practices (KAP) studies to inform tailored interventions. This study evaluated patients' KAP concerning lifestyle-related dry eye at the Ophthalmology Hospital of the China Academy of Chinese Medical Sciences from July 1 to July 26, 2024. A self-designed questionnaire was used to gather demographic data and assess KAP scores, yielding 556 valid responses (98.93%). Among participants, 342 (61.51%) were female, with a mean age of 39.26 ± 13.56 years. Mean KAP scores were 7.44 ± 4.65 (knowledge), 35.20 ± 4.10 (attitude), and 35.77 ± 6.15 (practice). Mediation analysis indicated direct influences of age and familial dry eye history on knowledge, with knowledge impacting attitude, and various factors influencing practice. While patients exhibited limited knowledge, they generally held positive attitudes and engaged in proactive practices. Enhancing patient education on lifestyle factors related to dry eye is essential to boost knowledge and foster effective prevention and management strategies.

Explore the factors related to the death of offspring under age five and appraise the hazard of child mortality using machine learning techniques in Bangladesh

BACKGROUND

Child mortality is a reliable and significant indicator of a nation's health. Although the child mortality rate in Bangladesh is declining over time, it still needs to drop even more in order to meet the Sustainable Development Goals (SDGs). Machine Learning models are one of the best tools for making more accurate and efficient forecasts and gaining in-depth knowledge. A deeper understanding is crucial for significantly reducing child mortality rates. Accurate predictions using machine learning models can empower authorities to implement timely interventions and raise awareness. So, the study aimed to explore the factors related to child mortality and assess the efficacy of various machine-learning models in predicting child mortality in Bangladesh.

METHODS AND MATERIALS

About Forty-two thousand observations, except the missing observations, were extracted for this study from the Bangladesh Demographic and Health Survey (BDHS) data conducted in 2017-18. The survey utilized a two-stage stratified sampling method, selecting 675 enumeration areas-250 in urban settings and 425 in rural areas-resulting in effective data collection from 672 clusters and 20160 households. The Chi-square test and recursive feature elimination (RFE) are used to find the relevant risk factors of child mortality among the number of factors. Six ML-based algorithms were implemented for predicting child mortality, such as Naïve Bayes, Classification and Regression Trees, Random Forest, C5.0 Classification, Gradient Boosting Machine, and Logistic Regression. Model evaluation metrics like accuracy, specificity, sensitivity, negative predictive value, F 1 score, positive predictive value, k-fold cross-validation, and area under the curve (AUC) techniques were used to evaluate the performance of the models.

RESULTS AND DISCUSSION

The child mortality rate is 8.2%, according to the data. The bivariate analysis showed that the child mortality rate was higher among the children whose mothers were uneducated, impoverished, underweight, aged 35-49, and gave birth before age 20. Families' water sources and religious connections had no statistically significant impact on child mortality. The prediction of child mortality using machine learning models is the main objective of this study. None of the machine learning models correctly classified dead occurrences. Therefore, this study conducted over-sampling and under-sampling analysis. Approximately 76727 and 6910 observations were sampled for over-sampling and under-sampling techniques, respectively. According to the findings of the over-sampling data, the Random Forest outperformed all the other models in terms of total performance based on training and testing sets, with an accuracy of seventy percent. The k-fold cross-validation approach demonstrated the Random Forest model's superior performance, and achieved the highest AUC (0.701). On the other hand, the Gradient Boosting Machine has the highest assessment for predicting child mortality in under-sampling analysis. The k-fold cross-validation also illustrated the better performance of the Gradient Boosting Machine.

CONCLUSION

The Gradient Boosting Machine and Random Forest produce the best predictive power for classifying child mortality and may help to ameliorate policy decision-making in this regard.

Bridging Data Gaps: Predicting Sub-national Maternal Mortality Rates in Kenya Using Machine Learning Models

Introduction Maternal mortality remains a critical global health issue, with ongoing efforts to reduce its incidence as part of international health priorities. Kenya, a sub-Saharan country that has a disproportionate number of maternal mortality is likely to miss this target unless evidence-based interventions are deployed. The paucity of reliable maternal health data calls for the development of alternative predictive models to complement the impaired civil registration system and the aperiodic national surveys. Methods We utilized DHS surveys from several Sub-Saharan African countries to estimate parameters for predicting Kenya's maternal mortality rate (MMR) in the absence of recent Kenya Demographic and Health Survey (KDHS) data. We developed a multiple linear regression model using supervised machine learning using the R-programming suite. Our model leverages machine learning techniques to analyze regional trends and predict sub-national MMR variations. We then applied the model to predict MMR for Kenyan counties using the data for the KDHS 2022 survey.  Results Using Pearson's correlation, we observed a significant positive correlation between MMR and total fertility (r = 0.32, p = 0.025) and a significant negative correlation between MMR and maternal age at first birth (r = -0.40, p = 0.005). Additionally, a significant correlation was observed with the cumulative percentage of mothers attending post-natal clinics, the prevalence of thinness (r = 0.77, p < 0.001), HIV infection in women (r = 0.20, p = 0.164), and physical violence during pregnancy. The model estimate of national MMR in 2022 was 367 deaths per 100,000 live births, ranging from 49 deaths per 100,000 live births in Kisii County to 1794 deaths per 100,000 live births in Turkana County. Conclusion Although MMR in Kenya displayed a general downward trend, our model's estimates for DHS 2022 indicate an increase compared to the 2019 National Census and Housing Survey estimate of 355 deaths per 100,000 live births. This rise may be attributed to COVID-19-related maternal deaths during the same period. The integration of predictive models to inform interventions and resource allocation could play a crucial role in enhancing maternal healthcare outcomes in Kenya.

Advances in AI and machine learning for predictive medicine

The field of omics, driven by advances in high-throughput sequencing, faces a data explosion. This abundance of data offers unprecedented opportunities for predictive modeling in precision medicine, but also presents formidable challenges in data analysis and interpretation. Traditional machine learning (ML) techniques have been partly successful in generating predictive models for omics analysis but exhibit limitations in handling potential relationships within the data for more accurate prediction. This review explores a revolutionary shift in predictive modeling through the application of deep learning (DL), specifically convolutional neural networks (CNNs). Using transformation methods such as DeepInsight, omics data with independent variables in tabular (table-like, including vector) form can be turned into image-like representations, enabling CNNs to capture latent features effectively. This approach not only enhances predictive power but also leverages transfer learning, reducing computational time, and improving performance. However, integrating CNNs in predictive omics data analysis is not without challenges, including issues related to model interpretability, data heterogeneity, and data size. Addressing these challenges requires a multidisciplinary approach, involving collaborations between ML experts, bioinformatics researchers, biologists, and medical doctors. This review illuminates these complexities and charts a course for future research to unlock the full predictive potential of CNNs in omics data analysis and related fields.

Henriques T. Machine learning models based on clinical indices and cardiotocographic features for discriminating asphyxia fetuses—Porto retrospective intrapartum study

Introduction

Perinatal asphyxia is one of the most frequent causes of neonatal mortality, affecting approximately four million newborns worldwide each year and causing the death of one million individuals. One of the main reasons for these high incidences is the lack of consensual methods of early diagnosis for this pathology. Estimating risk-appropriate health care for mother and baby is essential for increasing the quality of the health care system. Thus, it is necessary to investigate models that improve the prediction of perinatal asphyxia. Access to the cardiotocographic signals (CTGs) in conjunction with various clinical parameters can be crucial for the development of a successful model.

Objectives

This exploratory work aims to develop predictive models of perinatal asphyxia based on clinical parameters and fetal heart rate (fHR) indices.

Methods

Single gestations data from a retrospective unicentric study from Centro Hospitalar e Universitário do Porto de São João (CHUSJ) between 2010 and 2018 was probed. The CTGs were acquired and analyzed by Omniview-SisPorto, estimating several fHR features. The clinical variables were obtained from the electronic clinical records stored by ObsCare. Entropy and compression characterized the complexity of the fHR time series. These variables' contribution to the prediction of asphyxia perinatal was probed by binary logistic regression (BLR) and Naive-Bayes (NB) models.

Results

The data consisted of 517 cases, with 15 pathological cases. The asphyxia prediction models showed promising results, with an area under the receiver operator characteristic curve (AUC) &gt;70%. In NB approaches, the best models combined clinical and SisPorto features. The best model was the univariate BLR with the variable compression ratio scale 2 (CR2) and an AUC of 94.93% [94.55; 95.31%].

Conclusion

Both BLR and Bayesian models have advantages and disadvantages. The model with the best performance predicting perinatal asphyxia was the univariate BLR with the CR2 variable, demonstrating the importance of non-linear indices in perinatal asphyxia detection. Future studies should explore decision support systems to detect sepsis, including clinical and CTGs features (linear and non-linear).