Data-driven Approach to Detect and Predict Adverse Drug Reactions.

Area of Science:

• Pharmacovigilance and Computational Drug Safety
• Biostatistics and Machine Learning in Healthcare

Background:

• Adverse drug reactions (ADRs) are caused by a complex interplay of pharmacological, genetic, social, and patient-specific factors.
• Rapid advancements in data analysis methods, including machine learning and data mining, offer enhanced capabilities for ADR detection and prediction.
• Keeping pace with the evolving research landscape in ADR detection and prediction presents a significant challenge.

Purpose of the Study:

• To systematically review and synthesize recent research on adverse drug reaction (ADR) detection and prediction.
• To provide a consolidated overview of the methodologies and datasets employed in ADR research.
• To offer a valuable resource for researchers and pharmacists navigating the field of ADRs.

Main Methods:

• A comprehensive literature search was conducted to collect articles on ADRs published over the last twenty years.
• Articles were categorized based on the types of data utilized: omics, social media, and electronic medical records (EMRs).
• A review was performed focusing on the specific problems addressed, datasets used, and methodologies applied in each study.

Main Results:

• Three summary tables were created, detailing key information on ADR detection and prediction research.
• The review highlights the increasing reliance on diverse data sources for ADR analysis.
• The effectiveness of various statistical and machine learning approaches in ADR research is presented.

Conclusions:

• Data-driven approaches are highly effective for the detection and prediction of adverse drug reactions.
• This review offers a concise summary of the current state of ADR detection and prediction, aiding researchers and pharmacists.
• The findings underscore the importance of advanced analytical techniques in improving drug safety.