TY - JOUR
T1 - Artificial intelligence in sepsis early prediction and diagnosis using unstructured data in healthcare
AU - Goh, Kim Huat
AU - Wang, Le
AU - Yeow, Adrian Yong Kwang
AU - Poh, Hermione
AU - Li, Ke
AU - Yeow, Joannas Jie Lin
AU - Tan, Gamaliel Yu Heng
PY - 2021
Y1 - 2021
N2 - Sepsis is a leading cause of death in hospitals. Early prediction and diagnosis of sepsis, which is critical in reducing mortality, is challenging as many of its signs and symptoms are similar to other less critical conditions. We develop an artificial intelligence algorithm, SERA algorithm, which uses both structured data and unstructured clinical notes to predict and diagnose sepsis. We test this algorithm with independent, clinical notes and achieve high predictive accuracy 12 hours before the onset of sepsis (AUC 0.94, sensitivity 0.87 and specificity 0.87). We compare the SERA algorithm against physician predictions and show the algorithm’s potential to increase the early detection of sepsis by up to 32% and reduce false positives by up to 17%. Mining unstructured clinical notes is shown to improve the algorithm’s accuracy compared to using only clinical measures for early warning 12 to 48 hours before the onset of sepsis.
AB - Sepsis is a leading cause of death in hospitals. Early prediction and diagnosis of sepsis, which is critical in reducing mortality, is challenging as many of its signs and symptoms are similar to other less critical conditions. We develop an artificial intelligence algorithm, SERA algorithm, which uses both structured data and unstructured clinical notes to predict and diagnose sepsis. We test this algorithm with independent, clinical notes and achieve high predictive accuracy 12 hours before the onset of sepsis (AUC 0.94, sensitivity 0.87 and specificity 0.87). We compare the SERA algorithm against physician predictions and show the algorithm’s potential to increase the early detection of sepsis by up to 32% and reduce false positives by up to 17%. Mining unstructured clinical notes is shown to improve the algorithm’s accuracy compared to using only clinical measures for early warning 12 to 48 hours before the onset of sepsis.
UR - http://www.scopus.com/inward/record.url?scp=85100110345&partnerID=8YFLogxK
UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85100110345&origin=recordpage
U2 - 10.1038/s41467-021-20910-4
DO - 10.1038/s41467-021-20910-4
M3 - RGC 21 - Publication in refereed journal
C2 - 33514699
SN - 2041-1723
VL - 12
JO - Nature Communications
JF - Nature Communications
M1 - 711
ER -