Dynamic prediction of hospital admission with medical claim data

Table 2 Features pool used in predictive modeling

Feature Category	Features
Demographics	Age, Gender, Race.
Socioeconomics	Medicare status code, Beneficiary Dual Status code
Chronic condition^a	Any selected chronic conditions¹; Count of selected chronic conditions¹; Charlson Index Score [11]
Health care service^b	Count of a specific health care service utilization, including ED visit³, inpatient admission³, SNF stay, HHA stay and outpatient physician visit.
Acute exacerbation record^b	Count of ED visit or inpatient admission with selected exacerbation conditions².
DME utilization^b	Any DME usage; any oxygen-related DME usage.
Disease-specific procedure and service^c	Any cardio echo test; any spirometry test; any general pulmonary function test.
Medication^d	Count of unique prescription.
Location^e	Most recent care location prior to admission, including home, HHA, SNF, Inpatient and Outpatient
Cost^c	Total Expenditure

¹See Additional file 1: Table S1 for chronic conditions used in CHF predictive models
²See Additional file 1: Table S2 for exacerbation conditions used in CHF predictive models
³We included both the all-cause and disease-specific ED visit/inpatient admission
^aSuch features were collected during the 36 months window prior to the index date
^bSuch features were collected during the 6 months window prior to the index date
^cSuch features were collected during the 12 months window prior to the index date
^dSuch features were collected during the 3 months window prior to the index date
^eSuch features were collected during the 1 month window prior to the index date