Prediction of preterm and low birth weight risk using a physiology based artificial neural network integrating hematological, dental, and periodontal index markers: a cross sectional study based on machine learning

利用基于生理学的神经网络,结合血液学、牙科和牙周指标,预测早产和低出生体重风险:一项基于机器学习的横断面研究

阅读:1

Abstract

BACKGROUND: Preterm low birth weight (PLBW) is a major global contributor to neonatal morbidity and mortality. This study aims to develop and evaluate a novel artificial neural network (ANN) based predictive model that integrates hematological, periodontal, obstetric, and behavioral parameters to estimate the risk of PLBW with high accuracy. METHODS: A prospective case control study was conducted using data from 100 pregnant women, comprising 26 input parameters from diverse domains such as hematology (mean platelet volume [MPV], neutrophil to lymphocyte ratio [NLR]), periodontal status (Plaque Index [PI], clinical attachment loss [CAL]), obstetric history, and lifestyle behaviors (smoking, oral hygiene). A multilayer perceptron ANN architecture optimized by the Levenberg–Marquardt algorithm was employed to train the model and assess its predictive performance. RESULTS: The developed ANN model showed acceptable predictive performance in estimating PLBW risk and performed better than the logistic regression model in this dataset. The model effectively captured complex, nonlinear interactions among the input variables, with a mean squared error (MSE) of 0.1022 and a correlation coefficient (R) of 0.776, indicating good model fit. CONCLUSION: This study presents an interdisciplinary and scalable AI driven framework that leverages hematological, dental and periodontal, and behavioral markers for early prediction of PLBW. The ANN model may provide a basis for future tools aimed at prenatal risk stratification, pending validation in larger, multicenter cohorts.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。