Transformation of an acetaminophen toxicity nomogram to an artificial neural network.
A computerized artificial neural network was built based on the Rumack-Matthew nomogram for treatment of acetaminophen toxicity. NeuralWorks Predict 1.0 was used to build the network. The training set consisted of two input parameters, plasma concentration (microg/ml) and time of ingestion, and an output. The output was a binary unit representing the recommended use of N-acetylcysteine for the prevention of acetaminophen hepatotoxicity. The predictions made by the neural network correlated well with the expected values (95%). Since these networks can be modified with additional data, the present network may be useful for optimizing the use of N-acetylcysteine for the prevention of acetaminophen hepatotoxicity.