Centrala begrepp
The authors propose a machine learning approach to identify risk factors impacting birth events, aiming to create a user-friendly mobile application for neonatologists.
Sammanfattning
The content discusses the importance of training healthcare personnel in newborn care due to the unpredictability of high-risk situations. It introduces a machine learning approach to analyze real data and develop a mobile application for better recognition and intervention planning. Key factors like maternal hypertension, gestational age, and ventilation at birth are highlighted as crucial predictors of outcomes. The study showcases Decision Trees and Bayesian Networks as effective models for classification and correlation analysis, providing insights into neonatal resuscitation strategies.
Statistik
Neonatal mortality in Italy: 1.7 deaths per 1000 births.
Approximately 90% of newborns breathe spontaneously without interventions.
5% require positive pressure ventilation at birth.
Intubation rates vary between 0.4% and 2%.
Less than 0.3% need chest compression after birth.
Machine learning model accuracy: 97.2%.
APGAR score ≤ 7 observed in 18 patients sampled at 1 minute of life.
Outcomes include respiratory distress, NICU transfers, brain ultrasounds, and NIV requirements post-birth.
Citat
"Due to the low number of 'pathological' newborns with APGAR ≤ 7, the dataset was heavily unbalanced."
"Our models show that adding interplaying risk factors increases the probability of unfavorable outcomes."