AI Case Study
University of Kansas Medical Center identify drug interactions causing adverse effects in patients using a new method increasing accuracy by 10%
Researchers at the University of Kansas Medical Center update
an incremental rule mining algorithm to discover drug combinations that have resulted in adverse reactions in patients. The new method (CARD) demonstrated an improvement in accuracy of 10% over the older method while also identifying fewer drug interactions as causing the adverse effects.
Public And Social Sector
Education And Academia
According to Nature, "bioinformatician Mei Liu and her team reported finding adverse events from drug–drug interactions after mining through data in the US Food and Drug Administration's Adverse Event Reporting System, or FAERS4." The researchers report in Science Direct that "[t]o efficiently identify causal relationships, we introduce the causal concept into association rule mining by developing a method called Causal Association Rule Discovery (CARD). The properties of V-structures in Bayesian Networks are utilized in the search for causal associations. To demonstrate feasibility, CARD is compared to the traditional association rule mining (AR) method in DDI identification."
From Science Direct: "Based on physician evaluation of 100 randomly selected higher-order associations generated by CARD and AR, CARD is demonstrated to be more accurate in identifying known drug interactions compared to AR, 20% vs. 10% respectively. Moreover, CARD yielded a lower number of drug combinations that are unknown to interact, i.e., 50% for CARD and 79% for AR." However, as noted in Nature, the researchers acknowledge that the "system still needs fine-tuning before it can be used reliably to make clinical decisions".
From Nature: "Liu and her team modified a pre-established AI algorithm known as association-rule discovery or mining (AR) to create another algorithm that allowed computers to identify drugs that had not only associations with negative symptoms, but also causative relationships with these adverse symptoms. The algorithm assigns values between 0 and 1 to causal relationships it identifies. The closer to 1 the value is, the stronger the likelihood that the drug combination causes a given".
From the Science Direct paper: "Instead of reconstructing a causal Bayesian network, we propose to use the properties of causal Bayesian network to derive important properties of causal association rules, and apply such properties to guild the search of causal association rules."
According to the Science Direct paper: "Each time a person uses a prescription medication, there is a potential for an adverse drug reaction (ADR). That potential increases with an increase in the number of concurrent medications being used. Between 2009–2012, about 47% of the United States population reported they had used at least one prescription medication in the past 30 days and almost 11% reported using at least five prescription medications in the same period.
Numerous signal detection algorithms were designed for identifying relationships between drugs and adverse events (AEs) in spontaneous reports. Despite the number of existing algorithms, all have drawbacks that limit their effectiveness, such as noisy results with diminishing accuracy and low robustness due to low signal-to-noise ratio, high dimensionality of the data, and limited sample sizes."
7,700 drugs and 11,600 adverse events from the US FDA's Adverse Event Reporting System, published between 2004-2012.