Combining adult with pediatric patient data to develop a clinical decision support tool intended for children: leveraging machine learning to model heterogeneity

Table 4 Top important variables from each model

variable rank	Ventilator		ICU
variable rank	Combined RF	Pediatric RF	Combined RF	Pediatric RF
1	Specialty	Specialty	CPT 33,361–33,496 Surgical Procedures on Aortic Valve	Specialty
2	Service	Service	CPT 33,510–33,536 Venous Grafting for Coronary Artery Bypass	Weight
3	CPT 69,990–69,990 Operating Microscope Procedures	Height	CPT 33,508–33,508 Endoscopy Surrounding Vein for Coronary Artery Bypass	Height
4	CPT 61,510–61,516 Craniectomy or Craniotomy Procedures	Weight	Marker for Cardiac Surgery	Service
5	CPT 33,361–33,496 Surgical Procedures on Aortic Valve	History of Cardiovascular Disease	Specialty	Age
6	Weight	CPT 33,608–33,681 Repair Procedures for Single Ventricle or Cardiac Anomalies	Weight	Previous ambulatory encounters
7	Age	Age	Service	Diuretics
8	Height	Previous ambulatory encounters	Height	CPT 33,608–33,681 Repair Procedures for Single Ventricle or Cardiac Anomalies
9	Previous ambulatory encounters	CPT 61,343–61,343 Craniectomy for Decompression	Age	History of Cardiovascular Disease
10	CPT 20,650–20,664 Introduction or Removal Procedures on Musculoskeletal System	CPT 61,760–61,793 Stereotaxis Procedures on Skull, Meninges, and Brain	Previous ambulatory encounters	Previous Hospital Encounters

ISSN: 1472-6947