Perm Polytechnic Researchers Advance AI Tool to Predict Bronchopulmonary Dysplasia in Premature Infants
Scientists at Perm National Research Polytechnic University have developed a predictive approach to assess the risk of chronic lung disease in premature newborns. The aim is to give clinicians a clearer early signal so care strategies can be adjusted before trouble arises. The method blends clinical measurements with machine learning to provide a risk score that can be monitored over time, helping teams in Russia and beyond tailor interventions for fragile infants.
Each year in Russia, about 10 percent of the roughly 1.4 million births occur ahead of schedule. Premature birth frequently leads to bronchopulmonary dysplasia, a condition affecting the growth and development of the lungs. The combination of immaturity and medical treatments forms a landscape where careful monitoring and timely decisions are crucial for the infants long term respiratory health.
Bronchopulmonary dysplasia is the most common chronic lung disease seen in very low birth weight newborns, including those born weighing as little as one kilogram. The immaturity of lung tissue is linked to oxygen exposure during artificial ventilation and several other factors that can worsen outcomes.
In the new study, researchers built models that consider multiple quantitative indicators from newborns. The inputs include birth weight, nutritional status, features of the ventilation system, and other clinical parameters captured early after birth. The approach treats these variables as an interconnected system rather than isolated measures, allowing a more nuanced risk estimation.
Data came from 76 children, of whom 56 were diagnosed with moderate or severe bronchopulmonary dysplasia while the remainder had mild disease or none at all. The team used this dataset to train and test the predictive model, providing a realistic assessment of how the tool may perform in real clinical settings.
The researchers report that the predictive program achieved an accuracy of 84.4 percent, a result that underscores its potential to support early clinical decisions for at risk newborns in diverse settings.
Looking ahead, the authors anticipate that future algorithms like this may alert doctors in advance about the likelihood of a child being born with bronchopulmonary dysplasia. Early warnings could guide decisions about ventilation strategies, nutrition plans, and the intensity of monitoring, potentially reducing the severity of the disease for vulnerable newborns.
Earlier research into preterm health has shown some indicators can be favorable for premature infants in certain respects, highlighting resilience within data that predictive tools should support rather than replace clinical judgment.
While the results are encouraging, translating the model into bedside practice will require ongoing validation across diverse patient groups and healthcare settings. Real world deployment will demand careful attention to data quality, integration with electronic records, and clear guidance for clinicians on how to respond to risk signals.
From a broader perspective, the work aligns with global efforts to harness artificial intelligence in neonatal care. In North America, researchers and clinicians are actively exploring similar models to refine early warning systems, improve outcomes for preterm infants, and reduce long term respiratory complications.
In summary, the Perm Polytechnic project adds to a growing body of knowledge on how data driven approaches support neonatal teams in predicting and managing bronchopulmonary dysplasia in premature babies.