Skip to main content

Table 2 Prediction of risk for nosocomial infection

From: CDSS-RM: a clinical decision support system reference model

Day of Hospital Stay and Data availability

Use of models built in vitro with “Day 2” variables

Dynamic training & testing

Day 0: Demographics, admission diagnosis

High reported precision & recall is not realistic: Risky decisions, potential negative impact on patient safety

Precision and recall ~  70% are realistic: Use predictions with a lot of caution

Day 1: Demographics, admission diagnosis, medications, lab results

High in vitro reported precision & recall is not realistic: Risky clinical decisions, negative impact on patient safety

Precision and recall ~  80% are realistic: Use predictions with some caution

Day 2: Demographics, admission diagnosis, medications, more lab results, primary diagnosis

High in vitro reported precision & recall is realistic

Precision & recall ~  85% are realistic