HCViewer: software and technology for quality control and processing raw mass data of preventive screening

Abstract

AbstractA prototype automated system for the analysis and processing raw data of mass population screening in the Russian health centers (HCs), software program HCViewer, is developed. In it, essential elements of big data, including data filtering (i.e., detection and removal of outliers and fraud cases), statistics and spatial visualization are implemented. Testing HCViewer on the datasets of knowingly reliable, unreliable, and partially reliable bioimpedance body composition data has shown high diagnostic specificity (95.2–98.9%) and sensitivity (94.5–99.2%) of the filtering criteria. Retrospective analysis of the HCs’ bioimpedance database revealed rapid growth in the proportion of fraud cases, from 5–7% in 2009 to 43–45% in 2013–2014 (with the correction for false alarm rate), and the high level of heterogeneity in data quality, so that 80.3% of the incorrect data were generated in 20% of the HCs. This indicates inefficient management of the Russian preventive screening system and suggests incomparability and, therefore, uselessness of the related health statistics from some regions without a preliminary filtering of the underlying raw data. Our results show the potential utility of HCViewer for the dynamical quality control of the preventive screening data, as well as for the online health monitoring and modelling potential effects of various control measures aimed at the prevention of non-communicable diseases. After application of the selection criteria, a cross-sectional 2009–2015 database of bioimpedance measurements of 1.27 million individuals aged 5–96 years was formed for the assessment of population health. On the whole, fraud prevention should become the responsibility of the Russian health care system to ensure an efficiency of its function.