Artificial intelligence in sepsis early prediction and diagnosis using unstructured data in healthcare

Sepsis is a leading cause of death in hospitals. Early prediction and diagnosis are crucial but challenging due to overlapping symptoms with less critical conditions. This study develops the SERA algorithm, an AI algorithm using structured data and unstructured clinical notes to predict and diagnose sepsis. The algorithm achieves high predictive accuracy 12 hours before sepsis onset (AUC 0.94, sensitivity 0.87, specificity 0.87). Compared to physician predictions, SERA potentially increases early detection by up to 32% and reduces false positives by up to 17%. Incorporating unstructured clinical notes significantly improves accuracy compared to using only structured data.

Kim Huat Goh, Le Wang, Adrian Yong Kwang Yeow, Hermione Poh, Ke Li, Joannas Jie Lin Yeow, Gamiel Yu Heng Tan