Machine learning can predict which babies will be born with low birth weights

Babies born with low birth weight (less than 2.5 kg) are 20 times more likely to die. They are also more likely to develop neurological and cardiovascular diseases, diabetes, and growth problems later in life. A study conducted by researchers at the University of São Paulo (USP) shows that machine learning models can predict these cases, enabling early and more effective interventions and preventing complications.

The research, which represents the first application of advanced machine learning algorithms for this purpose in the country, was based on data from 1,579 pregnant women monitored by the Araraquara population cohort in the interior of the state of São Paulo, Brazil. Supported by FAPESP, the work also serves as a counterpoint to most studies of this type that use data from countries in the Global North.

The researchers tested four machine learning algorithms: Random Forest, XGBoost, LightGBM, and CatBoost. XGBoost was the most effective at identifying high-risk pregnancies.

“The findings have a significant impact on clinical practice and public policy formulation, given that the use of artificial intelligence and machine learning can enable earlier interventions, helping to reduce the risks associated with low birth weight and improving maternal and child health,” says Patrícia Rondó, a professor at the University of São Paulo’s School of Public Health (FSP-USP) and the author of the study published in the journal BMC Pregnancy and Childbirth.

Low birth weight is a global health problem with links to medical factors, such as complications during pregnancy, as well as socioeconomic factors, including maternal age, education, and access to prenatal care.

However, although the use of machine learning algorithms to predict low birth weight is gaining ground globally, most studies have been conducted in high-income countries. This limits their applicability in regions such as Brazil and Latin America.

According to the authors, the technology could enable healthcare professionals to implement early interventions, such as nutritional supplementation, maternal education, increased prenatal consultations, and counseling on lifestyle changes. These interventions could reduce the impact of the problem on newborns.

Rondó is also the coordinator of a population study conducted in Araraquara that assessed the nutritional status and body composition of 2,000 pregnant women and their children from the fetal stage onwards. In addition to serving as a basis for evaluating predictive algorithms for low birth weight, the sample, which is representative of the city of Araraquara and the surrounding region, has enabled a series of studies on obesity and genetic, environmental, and epigenetic factors associated with disease.

“The Araraquara cohort offers a unique opportunity by providing clinical, socioeconomic, behavioral, and environmental data on a population with characteristics that differ from those of populations in the Global North, where most studies of this type are conducted,” says Audêncio Victor, a data scientist and the lead author of the study. Victor is also a FAPESP fellow and the study is the subject of his doctoral research in epidemiology at USP, with a sandwich period at the London School of Hygiene and Tropical Medicine at London University.

Factors such as maternal age, anthropometric variables, socioeconomic status, and access to prenatal care were identified as key determinants of low birth weight. “The risk factors are well-known in the literature, and a predictive model such as the one we tested is important for screening higher-risk cases that deserve greater attention during prenatal care. In addition, these are simple, low-cost variables that are routinely collected in health services, which makes the model applicable even in regions with limited resources,” says Victor.

The researchers also found that the model based on data from Araraquara works for the population of southeastern Brazil, including the state of São Paulo, but there are limitations. “To apply the models in the Amazon or in African countries, for example, we’d need to make adjustments so that they’d become truly predictive. Each population has its own specific characteristics, and the models need to be calibrated so that they’re truly predictive in different geographical and social contexts,” the researcher adds.

The article “Predicting low birth weight risks in pregnant women in Brazil using machine learning algorithms: data from the Araraquara cohort study” can be read at: bmcpregnancychildbirth.biomedcentral.com/articles/10.1186/s12884-025-07351-3.

About São Paulo Research Foundation (FAPESP)
The São Paulo Research Foundation (FAPESP) is a public institution with the mission of supporting scientific research in all fields of knowledge by awarding scholarships, fellowships and grants to investigators linked with higher education and research institutions in the State of São Paulo, Brazil. FAPESP is aware that the very best research can only be done by working with the best researchers internationally. Therefore, it has established partnerships with funding agencies, higher education, private companies, and research organizations in other countries known for the quality of their research and has been encouraging scientists funded by its grants to further develop their international collaboration. You can learn more about FAPESP at www.fapesp.br/en and visit FAPESP news agency at www.agencia.fapesp.br/en to keep updated with the latest scientific breakthroughs FAPESP helps achieve through its many programs, awards and research centers. You may also subscribe to FAPESP news agency at http://agencia.fapesp.br/subscribe.