Association Rule Mining for Identifying High-Risk Drug Combinations in Overdose Fatalities: A Comparative Analysis of Apriori and FP-Growth Algorithms

Accidental fatal overdose is more often seen where the use of several substances occurs concurrently, e.g., opioids and stimulants. Association rule mining techniques in this research have been used to determine the high-frequency and highly probable fatal pairs of drugs based on an overdosage death database. In research, use of both the Apriori and FP-Growth algorithms used to identify patterns, quantify the association by utilizing support, confidence, and lift measures. Results show strong affinity, particularly for Xylazine and Fentanyl, with 99% confidence and 1.48 lift, representing a large co-occurrence that occurs over chance. Though Apriori has greater computational cost, it worked flawlessly due to sparsity in the data. FP-Growth, for its part, demonstrated its advantage in scalability as well as pattern discovery when its optimization was enabled. Comparative analysis not only reflects on methodological robustness and limitations but also provides useful suggestions to public health officials, forensic examiners, and policymakers. Some recommended suggestions are developing AI-driven early warning systems, enhancing forensic monitoring, and using gamified learning to raise awareness among youth. The findings highlight the importance of evidence-informed responses to stem the burgeoning epidemic of polysubstance overdoses.