Introduction: Adverse Drug Reactions (ADRs) are among the common causes of emergency department visits. Due to the complexity of diagnosing these reactions and the vast volume of clinical data, employing modern techniques such as data mining and text mining can play a significant role in identifying epidemiological patterns and predicting drug-related complications. This study aims to review original research articles (excluding systematic and narrative reviews) that have utilized data mining and text mining methods to analyze and detect ADRs.
Methodology: This theoretical study adopts a narrative review approach. Selected research articles were gathered from reputable scientific databases, focusing on the application of data mining and text mining algorithms in the context of drug-related adverse events
Findings: Out of 28 reviewed articles, 14 met the inclusion criteria and were categorized in Table 1 based on author(s), year, objective, study type, methodology, and overall findings. These articles were grouped into four main categories: development of data/text mining algorithms, analytical model design, model comparison, and deep learning-based prediction of ADRs following comprehensive model evaluations.
Result: The results show that data mining (through drug interaction and drug interaction identification, integration of multiple data sources, multi-omic data analysis, unsupervised learning from groups and channels in the health domain in social networks, and predictive modeling) and text mining (through side effect identification and extraction, big data analysis, drug side effect prediction, and clinical decision support) are powerful tools in drug side effect analysis and can help improve drug care, increase patient safety, and reduce treatment costs.