The Uppsala Monitoring Centre, on behalf of the World Health Organization (WHO), maintains the largest database of adverse drug reactions (ADRs), containing nearly two million reports. To address the challenge of identifying new drug-ADR signals from this vast dataset, a Bayesian confidence propagation neural network (BCPNN) has been developed. This method is designed to be flexible, automated, and robust, capable of handling large datasets and incomplete information. The BCPNN uses information theory to identify drug-ADR combinations with strong associations compared to the general data. The technique has been tested on time scan examples, demonstrating its ability to detect early signals and avoid false positives. When applied to quarterly updates, the BCPNN identified 1004 suspected drug-ADR combinations with a 97.5% confidence level, including 53 potentially serious ADRs related to new drugs. The results indicate that the BCPNN can effectively enhance the detection of significant signals in the WHO database, complementing expert assessments and improving the efficiency of signal generation.The Uppsala Monitoring Centre, on behalf of the World Health Organization (WHO), maintains the largest database of adverse drug reactions (ADRs), containing nearly two million reports. To address the challenge of identifying new drug-ADR signals from this vast dataset, a Bayesian confidence propagation neural network (BCPNN) has been developed. This method is designed to be flexible, automated, and robust, capable of handling large datasets and incomplete information. The BCPNN uses information theory to identify drug-ADR combinations with strong associations compared to the general data. The technique has been tested on time scan examples, demonstrating its ability to detect early signals and avoid false positives. When applied to quarterly updates, the BCPNN identified 1004 suspected drug-ADR combinations with a 97.5% confidence level, including 53 potentially serious ADRs related to new drugs. The results indicate that the BCPNN can effectively enhance the detection of significant signals in the WHO database, complementing expert assessments and improving the efficiency of signal generation.