Dataset Information

Machine learning model combining features from algorithms with different analytical methodologies to detect laboratory-event-related adverse drug reaction signals.

ABSTRACT:

Background

The importance of identifying and evaluating adverse drug reactions (ADRs) has been widely recognized. Many studies have developed algorithms for ADR signal detection using electronic health record (EHR) data. In this study, we propose a machine learning (ML) model that enables accurate ADR signal detection by integrating features from existing algorithms based on inpatient EHR laboratory results.

Materials and methods

To construct an ADR reference dataset, we extracted known drug-laboratory event pairs represented by a laboratory test from the EU-SPC and SIDER databases. All possible drug-laboratory event pairs, except known ones, are considered unknown. To detect a known drug-laboratory event pair, three existing algorithms-CERT, CLEAR, and PACE-were applied to 21-year inpatient EHR data. We also constructed ML models (based on random forest, L1 regularized logistic regression, support vector machine, and a neural network) that use the intermediate products of the CERT, CLEAR, and PACE algorithms as inputs and determine whether a drug-laboratory event pair is associated. For performance comparison, we evaluated the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), F1-measure, and area under receiver operating characteristic (AUROC).

Results

All measures of ML models outperformed those of existing algorithms with sensitivity of 0.593-0.793, specificity of 0.619-0.796, NPV of 0.645-0.727, PPV of 0.680-0.777, F1-measure of 0.629-0.709, and AUROC of 0.737-0.816. Features related to change or distribution of shape were considered important for detecting ADR signals.

Conclusions

Improved performance of ML models indicated that applying our model to EHR data is feasible and promising for detecting more accurate and comprehensive ADR signals.

SUBMITTER: Jeong E

PROVIDER: S-EPMC6248973 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundAntipsychotics (APs) have been associated with risk of torsade de Pointes (TdP). This has important public health implications. Therefore, (a) we exploited the public FDA Adverse Event Reporting System (FAERS) to characterize their torsadogenic profile; (b) we collected drug utilization data from 12 European Countries to assess the population exposure over the 2005-2010 period.MethodsFAERS data (2004-2010) were analyzed based on the following criteria: (1) ? 4 cases of TdP/QT abnormalities; (2) Significant Reporting Odds Ratio, ROR [Lower Limit of the 95% confidence interval>1], for TdP/QT abnormalities, adjusted and stratified (Arizona CERT drugs as effect modifiers); (3) ? 4 cases of ventricular arrhythmia/sudden cardiac death (VA/SCD); (4) Significant ROR for VA/SCD; (5) Significant ROR, combined by aggregating TdP/QT abnormalities with VA and SCD. Torsadogenic signals were characterized in terms of signal strength: from Group A (very strong torsadogenic signal: all criteria fulfilled) to group E (unclear/uncertain signal: only 2/5 criteria). Consumption data were retrieved from 12 European Countries and expressed as defined daily doses per 1,000 inhabitants per day (DID).ResultsThirty-five antipsychotics met at least one criterium: 9 agents were classified in Group A (amisulpride, chlorpromazine, clozapine, cyamemazine, haloperidol, olanzapine, quetiapine, risperidone, ziprasidone). In 2010, the overall exposure to antipsychotics varied from 5.94 DID (Estonia) to 13.99 (France, 2009). Considerable increment of Group A agents was found in several Countries (+3.47 in France): the exposure to olanzapine increased across all Countries (+1.84 in France) and peaked 2.96 in Norway; cyamemazine was typically used only in France (2.81 in 2009). Among Group B drugs, levomepromazine peaked 3.78 (Serbia); fluphenazine 1.61 (Slovenia).ConclusionsThis parallel approach through spontaneous reporting and drug utilization analyses highlighted drug- and Country-specific scenarios requiring potential regulatory consideration: levomepromazine (Serbia), fluphenazine (Slovenia), olanzapine (across Europe), cyamemazine (France). This synergy should be encouraged to support future pharmacovigilance activities.