Early prediction of neonatal sepsis using machine learning algorithms

ORIGINAL ARTICLE

Early Prediction of Neonatal Sepsis using Machine Learning Algorithms

Lakshman Narayanaswamy¹, Arvind Kasaragod², Kishore Kumar R³, Jayalakshmi Srinivasan².

¹Ketaki solutions, Bangalore, Karnataka, India,
²Department of Pediatrics, Cloudnine Hospitals, Bangalore, Karnataka, India,
³Department of Paediatrics and Neonatology, Cloudnine Hospital, Jayanagar, Bangalore, Karnataka, India.

Show affiliations

Abstract
Objective: We aim to develop a machine learning model based on data from Indian hospitals to predict neonatal sepsis several hours ahead of the current diagnosis used by hospitals to determine sepsis.
Methods: Vital data of neonates from nine Bangalore hospitals ICU was collected. The data was collated and normalized and then used as features for training a LSTM machine learning model built to predict sepsis 12 hours ahead. To train the model, patients with culture positive were taken as sepsis true and the rest were taken as sepsis false.
Results: Two models, one with hourly data input of 12 hours and the other with hourly data input of 24 hours were built. Both models predicted sepsis 12 hours ahead of the last given hourly input. Receiver Operating characteristics, an important model performance metric for the 12 hours model was 0.9324 and the 24 hour model was 0.8703.
Conclusion: The performance metrics of the machine learning model built using Indian neonatal vitals data shows lots of promise in aiding physicians in early detection on neonatal sepsis. The model requires field testing and collaboration with hospitals to ensure reliable data for broader adoption in neonatal sepsis detection.

Why this article important?

This manuscript is important as it presents a practical, artificial intelligence (AI)-driven solution for the early prediction of neonatal sepsis, which is a leading cause of neonatal deaths in low- and middle-income countries (LMICs). Traditional culture-based diagnosis is often delayed, limiting timely intervention. By using a Long Short-Term Memory (LSTM) model trained on routinely collected intensive care unit (ICU) data, the study demonstrates that sepsis can be predicted up to 12 hours in advance. The model is designed for real-world use with commonly available features, making it scalable across varied healthcare settings. Early detection through this tool can support faster clinical decisions and potentially improve neonatal outcomes.

Summary of article

This study focuses on predicting neonatal sepsis, a major cause of neonatal mortality in India, using machine learning (ML). Traditional sepsis diagnosis through culture results is often delayed, limiting timely intervention. A Long Short-Term Memory (LSTM) model was developed to analyze hourly vital signs from NICU patients across nine CloudNine hospitals in Bangalore, India. Data from 120 sepsis-positive and 415 non-sepsis neonates was collected, cleaned, and anonymized for model training using Python and TensorFlow. The model was trained to predict sepsis 12 hours in advance, using 12- and 24-hour time step inputs, with better performance observed with the latter. Emphasis was placed on using features widely available in most NICUs to ensure real-world applicability. The model supports early clinical decisions and helps differentiate true sepsis from similar conditions. Initial results are promising, and further clinical validation will support the model's integration into hospital settings for timely, life-saving interventions.

READ FULL TEXT

Home

Articles & Issues

About

Contribute

Pediatric Oncall Journal View Article

Home

Partner Sites