RESUMO
Objective To assess the factors associated with occurrence of stroke in Xuzhou region on the basis of stroke registry database using data mining methods Relief F algorithm.Methods Five hundred and forty-six patients with acute cerebral infarction from June 2013 to June 2014 in Stroke Registry database and 546 healthy people at the same period in the region were chosen;their clinical data were collected and retrospectively analyzed.And the related data were collected,arranged and put into epidata database.Firstly,the data were normalized and converted to the number between 1 and 10.Relief F algorithm was used to analyze the weight of past histories and hematology index between healthy control group and stroke group.Matlab software was used to program and calculate,and the main program runs for 20 times.And then,the obtained results for each weighted average were summarized.Finally,the weight of past medical history and blood parameters in healthy control group and stroke group were obtained.Results Relief F algorithm was applied for data analysis:as compared with healthy controls,stoke group had higher weight of infarction (0.0353125) and history of drinking (0.01175),while not higher weight of history of hypertension,diabetes mellitus or transient ischemic attack;as compared with healthy controls,stoke group had higher weight of uric acid nitrogen level (0.0072),blood uric acid level (0.0071),cholesterol level (0.0067) and homocysteine level (0.0064),followed by high-density lipoprotein (0.0062),low density lipoprotein (0.0041) and triglyceride (0.0039).Conclusion Application of Relief F algorithm can excavate the closely related potential risk factors for stroke in stroke registration database.