An Electronic Medical Record System with Treatment Recommendations Based on Patient Similarity.
Yu Wang,Yu Tian,Li-Li Tian,Yang-Ming Qian,Jing-Song Li
DOI: https://doi.org/10.1007/s10916-015-0237-z
IF: 4.92
2015-01-01
Journal of Medical Systems
Abstract:As the core of health information technology (HIT), electronic medical record (EMR) systems have been changing to meet health care demands. To construct a new-generation EMR system framework with the capability of self-learning and real-time feedback, thus adding intelligence to the EMR system itself, this paper proposed a novel EMR system framework by constructing a direct pathway between the EMR workflow and EMR data. A prototype of this framework was implemented based on patient similarity learning. Patient diagnoses, demographic data, vital signs and structured lab test results were considered for similarity calculations. Real hospitalization data from 12,818 patients were substituted, and Precision @ Position measurements were used to validate self-learning performance. Our EMR system changed the way in which orders are placed by establishing recommendation order menu and shortcut applications. Two learning modes (EASY MODE and COMPLEX MODE) were provided, and the precision values @ position 5 of both modes were 0.7458 and 0.8792, respectively. The precision performance of COMPLEX MODE was better than that of EASY MODE (tested using a paired Wilcoxon-Mann-Whitney test, p < 0.001). Applying the proposed framework, the EMR data value was directly demonstrated in the clinical workflow, and intelligence was added to the EMR system, which could improve system usability, reliability and the physician's work efficiency. This self-learning mechanism is based on dynamic learning models and is not limited to a specific disease or clinical scenario, thus decreasing maintenance costs in real world applications and increasing its adaptability.