Bacteremia is a common and life-threatening condition. It has an incidence of 140 to 160 per 100,000 person-years in the United States. Since bacteremia has many presentations, it can be challenging to diagnose. Subsequently there are very few guidelines on when to order a blood culture in an emergency setting. Neural networks are a means of machine learning and are presently being used in medicine to aid in decision making. With the use of machine learning, 22 variables that have been associated with infection and bacteremia were used to build a neural network to determine which variables associated with bacteremia are most associated with a positive blood culture.
Giovane Richard, Sheppard Robert A
deep neural network, machine learning healthcare data, persistent bacteremia, positive blood culture, severe sepsis