Jump-start central statistical monitoring in your trial in minutes

  • Home
  • Jump-start central statistical monitoring in your trial in minutes
nadmin March 9, 2019 0 Comments

Risk based monitoring, a key component of centralized monitoring, is gaining popularity in recent years after encouragement from regulatory agencies. Often teams who adopt risk-based monitoring, limit its application to review of pre-defined site-level operational and clinical data summaries, and tracking of KRIs and KPIs (key Risk Indicators and Key performance Indicators).

In contrast to such monitoring methods, Central Statistical monitoring (CSM) is driven based on actual data collection. Although some of the CSM approaches can be quite complex, the concept behind such methods is simple. Data collected within a trial and across centers, countries or cohorts is homogenous and have common patterns. This is since such data is collected as part of data collection plan outlined in clinical protocol. Any deviations from this general behavior within a site, country or cohort could be due to deviations from study procedures, data entry errors, inaccurate measurements from poorly calibrated medical equipment, sloppiness, or fraud. Hence, it is important to evaluate such anomalies to optimize monitoring activities, detect potential data issues during the early phase of a trial and ensure high level of data quality.

Although, CSM has the potential to identify systemic issues and improve data quality, its adoption has been slow in the industry. The key reason for this is lack of a standard, cost-effective and open approach to implement CSM. Usual CSM approaches involve complex proprietary algorithms, costly technology and trial specific custom implementations. These approaches may not be tailored for short and smaller trials. Even after investing all the time and money in CSM, the findings can be hard to incorporate in typical clinical data workflow.

To overcome these challenges, we at Nimble Clinical Research have developed an open and free Anomaly Detection software. Its key features are:

  • Detect anomalies in unsupervised manner, i.e. without spending any time on data preparation, one can use the software to detect anomalies from CDISC standardized data in minutes.
  • Perform deeper learning of the data in order to find hidden anomalies beyond just flagging straightforward univariate extreme values
  • Use of sound, simple and predictable statistical approaches allowing teams with diverse backgrounds to readily understand the findings.
  • Applicable for small to large clinical trials
  • Findings are presented as an actionable list of anomalies that are easy to understand for cross-functional teams.
  • Achieve higher signal to noise ratio in findings to ensure the end users spend time mostly on high quality and interesting findings.
  • Visually highlight relative interestingness of the findings to bring attention to most interesting patterns

More details for the approach and the free copy of the software can be requested from info@nimble-cr.com. Feel free to reach me directly (email: vineet.jain@nimble-cr.com) for questions or support.