We offer a simple yet powerful clinical data quality framework for anomaly detection. It uses formal statistical tests and unsupervised machine learning techniques to find anomalies in clinical data. The increasing computational abilities of today’s systems and adoption of CDISC data standards across industry makes it easier to apply this framework. With availability of standardized source data, the approach can be applied on most of the clinical studies with little effort.
To know more, check our paper presented at PhUSE US Connect 2019.