Abstract

PV SIGNAL DETECTION USING STATISTICAL DATA MINING METHODS

*Atharva Bharat Karale and Dr. M. D. Game

Abstract

Pharmacovigilance programmes monitor and help safeguarding the use of medicines which is grave to the success of public health programmes. Identifying new possible risks and developing risk minimization action plans to prevent or ease these risks is at the heart of all pharmacovigilance activities throughout the product lifecycle. In this paper we examine the use of data mining algorithms to identify signals from adverse events reported. The capabilities include screening, data mining and frequency tabulation for potential signals, including signal estimation using established statistical signal detection methods. We have standard processes, algorithms and follow current requirements for signal detection and risk management activities. The Safety Evaluators, who are familiar with the current labeling, known adverse events, and mechanism of actions of their drugs, read the reports, look for particular abnormalities or issues relative to the normal product safety profile, and check the validity of the report. If the collection of reports is regarded important due to abnormalities or issues after this process, the drug and adverse event relationship is investigated more thoroughly and regulatory action may be taken. In this paper various statistical data mining algorithms and statistical analyses used to find patterns within sets of data at the FDA. With data mining, the FDA can improve its report analysis process by automatically selecting the most significant reports for review as well as allowing reviewers to view the information from all the reports received in an organized manner, instead of having to manually consider each one. The reports that may contain serious and unexpected adverse events.

Keywords: Adverse drug reactions, pharmacovigilance, safety signals, statistical methods.