Skip to main content

Table 5 Ranking of key features with optimal accuracy for surgical expenses

From: Machine learning-driven prediction of medical expenses in triple-vessel PCI patients using feature selection

Rank

Variables averaged through LR, RF, SVR, GLMBoost, BayesGLM and eXGB

Rank

Variables selected by eXGB

One year before the variables

 1

【X10】Hospitalization expenditure

1

【X10】Hospitalization expenditure

 2

【X8】Mechanical ventilation

2

【X9】Outpatient expenditure

 3

【X3】CCI scores

3

【X2】Age

 4

【X9】Outpatient expenditure

4

【X8】Mechanical ventilation

 5

【X7】Blood transfusion

5

【X3】CCI scores

 6

【X2】Age

6

【X7】Blood transfusion

 7

【X6】ED visits

7

【X6】ED visits

 8

【X59】Anticoagulant

8

【X21】Septicemia

 9

【X44】Dopamine

9

【X20】Asthma

 10

【X52】Lipid lowering

10

【X44】Dopamine

 11

【X18】COPD

11

【X59】Anticoagulant

 12

【X28】Dementia

12

【X48】DPP4i

 13

【X56】B-blocker

13

【X37】Major bleeding

 14

【X40】LRTI

14

【X31】AF

 15

【X37】Major bleeding

15

【X19】Gout

 16

【X31】AF

16

【X62】Uric acid

 17

【X48】DPP4i

17

【X40】LRTI

 18

【X14】Hyperlipidemia

18

【X36】Gastrointestinal bleeding

 19

【X25】AKF

19

【X52】Lipid lowering

 20

【X57】Insulin

20

【X55】Alpha-blockers

 21

【X53】Sulfonylureas

21

【X28】Dementia

 22

【X63】Acarbose

22

【X23】Palpitations

 23

【X60】TZD

23

【X16】DM

 24

【X16】DM

24

【X63】Acarbose

 25

【X49】NASID

25

【X22】Peptic ulcer

 

26

【X39】Intraabdominal

27

【X25】AKF

28

【X60】TZD

  1. Abbreviations CCIs Charlson comorbidity index score, ED Emergency department, DM diabetes mellitus, CKD Chronic kidney disease, COPD Chronic obstructive pulmonary disease, AKF Acute kidney failure, AF Atrial fibrillation, LRTI lower respiratory tract infections, DPP4i Dipeptidyl peptidase 4 inhibitor, NASIDs Non-steroidal anti-inflammatory drugs, TZD Thiazolidinedione