Keywords

Preterm birth; Computer models; Machine learning

Introduction

Premature birth (PTB) is a severe public health issue that has negative consequences for both families and society. It is the largest cause of neonatal mortality and morbidity worldwide, as well as the second leading cause of death in children under the age of five. PTB has been a key research topic in the healthcare area for the past two decades. Pregnancy and childbirth opened the way for medical professionals and researchers to investigate several efficient strategies for preventing preterm birth in women who are experiencing pregnancy-related difficulties [1]. These measures include providing healthcare services to all pregnant women in order to control PTB, as well as any medical interventions targeted at increasing women's awareness of early signs of pregnancy problems.

A pregnant woman's maternal history is an important aspect of neonatal studies for delivering various clinical therapies to newborn babies' health, disease, care, and results. Newborn babies are unique in that they have no prior medical history and their early neonatal path is directly linked to their mothers' maternal history [2]. Arrangements for critical social and economic support for women before, during, and after pregnancy, including educational, medical, and other training programmes that facilitate healthy parenthood, are also included in the healthcare services. In general, physicians treat disorders (including PTB) depending on their understanding. However, because medical knowledge varies from expert to expert, manual diagnosis may not always be correct.

Manual treatment, on the other hand, is a time-consuming task. Furthermore, the demand for medical specialists is growing every day as the world's population grows, and in developing nations such as India, a huge percentage of women come from low- and middle-income homes. They do not have access to adequate healthcare or health education to learn about any complications that may emerge during pregnancy, particularly in rural areas. Furthermore, people are generally wary of physicians' prescriptions since doctors frequently mislead patients by recommending expensive tests (such as the double marker test, foetal echocardiogram, urine test, and FT4 test, which are used to diagnose any pregnancy issues) [3]. In addition, doctor's appointments are often on the higher end of the spectrum. Furthermore, clinicians may occasionally diagnose cases incorrectly.

Preterm delivery is, after all, the most serious problem in gynaecology and obstetrics, as well as a major health risk for all pregnant women. For diagnosing high-risk patients, it may be necessary to do many ultrasound sonography (USG) tests in addition to paying a doctor's appointment fee, all of which might add up to a significant expenditure that is beyond the means of many families. As a result, developing a computerised system (i.e., an e-healthcare system) for birth prediction based on previous diagnosis data is a critical answer for making prompt and accurate decisions in the event of a bad pregnancy outcome, saving lives and money. Because of advances in technology and the digitization of medical records, the Obstetrical community is undergoing a pioneering transformation.

One of the most promising instruments for medical research and development is data analytics [4]. Machine learning techniques (such as neural networks, support vector machines, logistic regression, and Decision Trees) are now being used to construct disease predicting models in order to meet the expanding demands of human experts in the medical field. Medical datasets, on the other hand, are highly unbalanced, contradictory, and ambiguous. As a result, developing an effective intelligent model for medical datasets is a difficult undertaking. One such clinical dataset is the PTB dataset. For the prediction of PTB, researchers have developed a number of predictive models based on standard intelligent methods. They do, however, have a number of flaws, such as a lack of understand ability and inefficiency in making rapid and precise decisions.

Furthermore, early detection and diagnosis are crucial in preventing such problems. Early detection of such diseases may be aided by machine intelligence models based on symptoms. The longer it takes to receive a clinical decision for preterm birth, the higher the risk of pregnancy problems, which in turn raises the chance of perinatal mortality. Neonatal health is also very significant in the obstetrical community because of its direct link to prenatal mortality. PTB is responsible for 35% of newborn deaths, according to a UNICEF research published in 2015 [5]. Due to a lack of health facilities and a shortage of healthcare staff, the rate of PTB is rising in rural parts of most developing countries.

Conclusion

The mechanical integrity of the soft tissue components that support the foetus could have a role in keeping the pregnancy healthy and inducing labour. The amount of mechanical loading placed on the uterine, cervix, and foetal membranes during pregnancy is still unknown, but it is thought that this overstretching contributes to the premature initiation of contractility, tissue remodelling, and membrane rupture, which leads to preterm birth.

References

1. Bengio Y (2009) Learning Deep Architectures for AI; Now Publishers Inc: Norwell, MA, USA, 1-127.
2. Ward RM, Beachy JC (2003) Neonatal complications following preterm birth. BJOG Int J Obstet Gynaecol, 110; 8-16.
3. Tang Y, Zhang YQ, Chawla N, Krasser S (2008) SVMs Modeling for Highly Imbalanced Classification. IEEE Trans. Syst. Man Cybern. Part B Cybern, 39; 281-288.
4. Boulle M (2004) Khiops: A Statistical Discretization Method of Continuous Attributes,” Machine Learning, 55(1); 53-69.
5. Witten H, Frank E (2002) Data mining. ACM Sigmod Record, 31(1): 76-77.