Analyzing adverse drug reaction using statistical and machine learning methods

Abstract

Background: Adverse drug reactions (ADRs) are unintended negative drug-induced responses. Determining the association between drugs and ADRs is crucial, and several methods have been proposed to demonstrate this association. This systematic review aimed to examine the analytical tools by considering original articles that utilized statistical and machine learning methods for detecting ADRs. Methods: A systematic literature review was conducted based on articles published between 2015 and 2020. The keywords used were statistical, machine learning, and deep learning methods for detecting ADR signals. The study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement (PRISMA) guidelines. Results: We reviewed 72 articles, of which 51 and 21 addressed statistical and machine learning methods, respectively. Electronic medical record (EMR) data were exclusively analyzed using the regression method. For FDA Adverse Event Reporting System (FAERS) data, components of the disproportionality method were preferable. DrugBank was the most used database for machine learning. Other methods accounted for the highest and supervised methods accounted for the second highest. Conclusions: Using the 72 main articles, this review provides guidelines on which databases are frequently utilized and which analysis methods can be connected. For statistical analysis, >90% of the cases were analyzed by disproportionate or regression analysis with each spontaneous reporting system (SRS) data or electronic medical record (EMR) data; for machine learning research, however, there was a strong tendency to analyze various data combinations. Only half of the DrugBank database was occupied, and the k-nearest neighbor method accounted for the greatest proportion.

An evaluation of adverse drug reactions and outcomes attributed to kratom in the US Food and Drug Administration Adverse Event Reporting System from January 2004 through September 2021

“…Reports were deduplicated only via the automated deduplication process, as it was not feasible to visually inspect all FAERS reports. There is no gold standard for signal detection using spontaneous reports and each method has advantages, disadvantages, and limitations 24,25 . Two different disproportionality signal detection methods were used in sequence.…”

“…There is no gold standard for signal detection using spontaneous reports and each method has advantages, disadvantages, and limitations. 24 , 25 Two different disproportionality signal detection methods were used in sequence. First, reporting odds ratios were used to identify preliminary signals because the method tends to have a relatively high sensitivity.…”

Drugs That Interact With Colchicine Via Inhibition of Cytochrome P450 3A4 and P-Glycoprotein: A Signal Detection Analysis Using a Database of Spontaneously Reported Adverse Events (FAERS)

“…30 There is no gold standard for signal detection involving DDIs, and each method has advantages, disadvantages, and limitations. 31,32 Specific to this study, the ROR tends to be a sensitive but produces a greater number of false positives when there are small cell counts. Norén et al 33 have proposed methods that are more conservative (produce fewer false positive signals) by regularizing for unstable variance where there are small cell counts.…”

Reply to: Comments on “Adverse reactions associated with immune checkpoint inhibitors and bevacizumab: A pharmacovigilance analysis”

“…1 It is now widely accepted that there is no gold standard for monitoring adverse drug reaction (ADR) signals and that each monitoring method has strengths and weaknesses. 2 In his comment, Dr Noguchi pointed out the shortcomings of the reporting odds ratio (ROR) algorithm and recommended the use of the Ω shrinkage measure to verify ADR signals associated with the use of bevacizumab combined with immune checkpoint inhibitors (ICIs). The Ω shrinkage method proposed by Norén et al is an alternative method of efficiently calculating the observed-to-expected ratio (O/E) for detecting potential drugdrug interaction (DDI) signals in spontaneous reporting systems (SRS).…”

“…The Ω shrinkage method proposed by Norén et al is an alternative method of efficiently calculating the observed-to-expected ratio (O/E) for detecting potential drugdrug interaction (DDI) signals in spontaneous reporting systems (SRS). 2 This algorithm avoids the potential problem of model estimates of basic multiplicative risk that deviate from additive correlation by explicitly modeling additive risk. 3 Put another way, the Ω shrinkage measure effectively avoids an inflated ADR signal score caused by a small number of reports, and is considered to be the most conservative algorithm for monitoring ADR signals.…”