Sepsis is the primary cause of death from infection, especially if not recognized and treated promptly [25]. Its early diagnose is crucial so that appropriate treatment can be started promptly and patients can get the best possible chances of survival. This is especially true in Intensive Care Unit (ICU), where patients are seriously ill and need promptly treatments and diagnoses. For these reasons, there are crucial scores that are indicators of the disease, but they must be calculated after a period of time that may fall too much ahead of critical situations. Taking advantage of the studies underlying such scores, we propose a simple yet powerful machine learning framework for detecting sepsis in a period of few hours after the ICU admission. Moreover, by means of conformal prediction techniques [30], such framework avoids providing the physicians with the predicted value for the instances whose data cannot either confirm or exclude a Sepsis diagnosis in the current time frame. We extensively test our framework using the MIMIC-III database [12].

On the early detection of Sepsis in MIMIC-III

Medina, Manuel;Sala, Pietro
2021

Abstract

Sepsis is the primary cause of death from infection, especially if not recognized and treated promptly [25]. Its early diagnose is crucial so that appropriate treatment can be started promptly and patients can get the best possible chances of survival. This is especially true in Intensive Care Unit (ICU), where patients are seriously ill and need promptly treatments and diagnoses. For these reasons, there are crucial scores that are indicators of the disease, but they must be calculated after a period of time that may fall too much ahead of critical situations. Taking advantage of the studies underlying such scores, we propose a simple yet powerful machine learning framework for detecting sepsis in a period of few hours after the ICU admission. Moreover, by means of conformal prediction techniques [30], such framework avoids providing the physicians with the predicted value for the instances whose data cannot either confirm or exclude a Sepsis diagnosis in the current time frame. We extensively test our framework using the MIMIC-III database [12].
978-1-6654-0132-6
SOFA Score, MIMIC III, Conformal Prediction, Convolutional Neural Network
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11562/1052860
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact