Because electronic health records are often missing information about patients, we developed and validated a tunable computer algorithm to identify subsets of patients whose data are relatively complete and therefore better suited for clinical research studies. A manuscript describing this study is currently under review:
Griffin M Weber, William G. Adams, Elmer V. Bernstam, Jonathan P. Bickel, Kathe P. Fox, Keith Marsolo, Vijay A. Raghavan, Alexander Turchin, Xiaobo Zhou, Shawn N. Murphy, Kenneth D. Mandl. “A Tunable Approach to Identifying Research Cohorts with ‘Complete Data’ in Electronic Health Records”. 2015. Under Review.
Run the following database script on your i2b2 CRC cell's database (Microsoft SQL Server) to identify patients whose data are relatively complete.
For more information, please contact
Griffin M Weber, MD, PhD
Department of Biomedical Informatics
Harvard Medical School