Evaluation and analysis of patterns and relationships among data in Hospital Information System (HIS) at Marand Razi Hospital using data mining

Abstract

Abstract Introduction: Hospital information system is one of the rich data resources. If their data are analyzed through data mining, it can bring invaluable results to planning and decision making for quality service delivery. Aim: Aim of this research was to identify hidden patterns and associations among data stored in hospital information system. Methods: This research is a descriptive and analytical study of modeling type that has been carried out during March 2011- March 2014. Data collection was conducted through taking reports from hospital information system. Data of all patients (n=33605) hospitalized during period of study were included in this research. After preliminary preprocessing, data was analyzed and evaluated through using Rapid Miner and SPSS. Results: Findings on pattern of patient entry to hospital revealed that hospitalization due to pregnancy, child birth and the puerperium had the highest frequency (32.8%) followed by those hospitalized due to cardiovascular diseases (13%). The lowest number of hospitalization was due to mental and behavioral disorders. Hospitalization trend of patients was consistent over 3 years. Cardiovascular disease was found as main cause of death among cases. This category had also the highest cost compared with other type of diseases. Patient hospitalized with endocrine, metabolic and Nutritional diseases had longest average of LOS. The shortest average of LOS was observed among patients diagnosed with diseases of the eye and adnexa. In addition decision tree for factors effecting LOS, cost and patient death is presented and evaluated. Conclusion: Data mining can reveal hidden patterns and associations in data residing in hospital information system. Key words: data mining, visualization, length of stay, cost, mortality