How do you define a hemodynamically significant PDA when the definition itself is part of the problem? Dr. Giulia Lima is a third year fellow at Harvard who built a machine learning model to predict hemodynamic significance in premature babies with PDA using continuous physiological data: heart rate variability, diastolic blood pressure, and type of respiratory support. Gestational age, the most statistically significant variable in univariable analysis, did not make the final model. The model reached over 90% sensitivity, which means if it says there is no hemodynamically significant PDA, you can probably trust that. The model is only as good as the labels you give it. And right now, we are not always giving it the right ones. Every second at the bedside, you are generating data your brain cannot fully process. This episode is about what happens when you build a system that can. Giulia also shares what three years at Harvard taught her about mentorship, saying yes, and knowing when the program knows better than you do. 🎧 Listen now on Apple Podcasts: buff.ly/CcDUiNU #nicu #neonatal #babies #medicine #newborn #innovation #podcast