RESUMO
Objective:To explore a solution for the construction of the CCI index with an example of cerebral infarction to provide a guide for adjusting the policy of Diagnosis-Intervention Packet.Methods:Lasso regression was used to screen for comorbidities to construct the model,K-means clustering was used for case severity categorization.CCI indices were calculated as one plus the sum of standardized regression coefficients.Results:According to the analysis,five key variables were found,including other disorders of the lungs,non-insulin-dependent diabetes mellitus with ketoacidosis,systemic inflammatory response syndrome of non-infectious etiology with organ failure,acute duodenal ulcer with bleeding,and chronic obstructive pulmonary disease with acute lower respiratory tract in-fection,with CCI indices of 1,1,1.026,1.034,and 1.101.Simulated calculation's result showed a decrease in medical insurance pay-ment losses after applying CCI indices.Conclusion:The CCI index construction scheme based on Lasso regression and K-means clus-tering is reasonable and effective.