Standardization and structuring of data are necessary for analysis of medical data collected from different medical in-stitutions. An electronic message and repository have been developed to link electronic medical records in this research project, which has simplified the integration of data. In this paper, we propose an analysis method to determine the prior-ity of clinical intervention by clustering, visualizing time se-ries, and prioritizing of patient outcome and patient status during hospitalization. Data analysis to long-term hospitali-zations was performed using reimbursement data and clinical pathway data stored in the institutions’ repositories using three versions of machine learning. We confirmed the distri-bution by clustering, tendency by the directed graph, extrac-tion of logically important variables, and effects by SHapley Additive Explanation. Constipation was determined to be the risk factor most strongly related to long-term hospitalization. Drainage management was identified as a factor that can improve long-term hospitalization. This study achieved to extract patient status s extraction to give feedback to Learn-ing Health System. We successfully generated evidence of med-ical processes by gathering patient status, medical purpose, and patient outcome with high data quality from multiple institutions, which were difficult with conventional electronic medical records.