Drug risk assessment method based on ADR monitoring report and outlier detection technology

Abstract

The invention provides a drug risk assessment method based on an ADR monitoring report and an outlier detection technology, and the method comprises the following steps: obtaining an original ADR database, and carrying out the normalization processing of drug names and adverse reaction names in the original database; determining outlier detection research objects and parameters: calculating a PRRvalue and an IC value of each drug-adverse reaction combination by adopting a proportional imbalance method, and constructing a spatial vector model by taking the PRR value and the IC value as features; determining a clustering number by utilizing an elbow rule, and finding a parameter k for minimizing a cost function; clustering by adopting a K-Means clustering algorithm, and randomly selecting kdata points as an initial centroid for clustering iteration according to a clustering number k determined by an elbow rule; calculating outlier scores of all the data points; and determining a threshold value and carrying out outlier detection on each cluster: setting the threshold value according to outlier score conditions, and determining points exceeding the threshold value as outliers. The invention provides an abnormal signal detection method for drug alert.

Description

technical field [0001] The invention relates to an evaluation method, in particular to a drug risk evaluation method, which belongs to the technical field of data mining. Background technique [0002] When drugs treat patients, they may also bring side effects unrelated to the drugs, which are called adverse drug reactions (ADR). Pharmacovigilance of marketed drugs is a very important task. At present, all countries in the world have established voluntary reporting monitoring systems and accumulated a large amount of ADR data resources. However, the use of these data focuses on the application research of traditional adverse reaction signal detection methods (PRR, IC, MHRA, etc.), mainly mining conventional signals, but lack of research on rare signal mining methods. These rare signals often show some distinctive characteristics, contain more important information, and have extremely important reference value for the hazard research of drugs. It is innovative and necessar...

AI Technical Summary

Problems solved by technology

However, the use of these data focuses on the application research of traditional adverse reaction signal detection methods (PRR, IC, MHRA, etc.), mainly mining conventional signals, but lack of research on rare signal mining methods

Images