Medical Student Research Forum Posters

• The Pediatric Emergency Care Applied Research Network (PECARN) conducted a study of ~42,000 children with minor blunt head trauma and developed and validated a clinical prediction rule to identify those at low risk of clinically-important traumatic brain injuries (ciTBIs) .1

• Prior studies have relied on traditional multivariable statistical methods,1-2 but more recent research regarding prediction rules has used machine learning (ML). 3-6

• In a previous study, investigators created a ML algorithm analyzing the PECARN dataset using a single decision tree that fits all nodes simultaneously, a complicated model at risk of over fitting. 6

• In this study, we created multiple algorithms (see Table 2) using ML for classification of children at risk for ciTBIs via the PECARN head trauma public use dataset. The model predictions were statistically compared to no information rates, the error rate when the input and output are independent.