Artificial neural networks are the main tools for data mining and were inspired by the human brain and nervous system. Studies have demonstrated their usefulness in medicine. However, no studies have used artificial neural networks for the prediction of adverse drug reactions. We aimed to validate the usefulness of artificial neural networks for the prediction of adverse drug reactions and focused on vancomycin -induced nephrotoxicity. For constructing an artificial neural network, a multilayer perceptron algorithm was employed. A 10-fold cross validation method was adopted for evaluating the resultant artificial neural network. In total, 1141 patients who received vancomycin at Hokkaido University Hospital from November 2011 to February 2019 were enrolled. Among these patients, 179 (15.7%) developed vancomycin -induced nephrotoxicity. The top three risk factors of vancomycin -induced nephrotoxicity which are relatively important in the artificial neural networks were average vancomycin trough concentration ≥ 13.0 mg/L and concomitant use of piperacillin–tazobactam and vasopressor drugs. The predictive accuracy of the artificial neural network was 86.3% and that of the multiple logistic regression model (conventional statistical method) was 85.1%. Moreover, area under the receiver operating characteristic curve (AUROC) of the artificial neural network was 0.83. In the 10-fold cross-validation, the accuracy obtained was 86.0% and AUROC was 0.82. The artificial neural network model predicting the vancomycin -induced nephrotoxicity showed good predictive performance. This appears to be the first report of the usefulness of artificial neural networks for an adverse drug reactions risk prediction model.

We previously constructed a risk prediction model of vancomycin (VCM)-associated nephrotoxicity for use when performing initial therapeutic drug monitoring (TDM), using decision tree analysis. However, we could not build a model to be used at the time of initial administration due to insufficient sample size. Therefore, we performed a multicenter study at four hospitals in Japan. We investigated patients who received VCM intravenously at a standard dose from the first day until the initial TDM from November 2011 to March 2019. Acute kidney injury (AKI) was defined according to the criteria established by the “Kidney disease: Improving global outcomes” group. We extracted potential risk factors that could be evaluated on the day of initial administration and constructed a flowchart using a chi-squared automatic interaction detection algorithm. Among 843 patients, 115 (13.6%) developed AKI. The flowchart comprised three splitting variables (concomitant drugs (vasopressor drugs and tazobactam/piperacillin) and body mass index ≥ 30) and four subgroups. The incidence rates of AKI ranged from 9.34 to 36.8%, and they were classified as low-, intermediate-, and high-risk groups. The accuracy of flowchart was judged appropriate (86.4%). We successfully constructed a simple flowchart predicting VCM-induced AKI to be used when starting VCM administration.

Therapeutic drug monitoring for voriconazole, an antifungal agent, is essential for maximizing efficacy and preventing toxicity. The aim of this study was to elucidate the optimal maintenance dose of voriconazole in patients with severe liver cirrhosis (Child–Pugh class C) by reviewing the plasma trough concentrations obtained by therapeutic drug monitoring and daily doses of voriconazole. We retrospectively evaluated 6 patients with Child–Pugh class C cirrhosis who received oral voriconazole treatment and were liver transplant recipients or were awaiting liver transplantation. We compared their voriconazole trough concentrations and daily maintenance doses to those of patients who did not have liver cirrhosis (n=56). We found that plasma voriconazole trough concentrations in all patients with Child–Pugh class C were almost within therapeutic range, and the median plasma trough concentration at steady state was not significantly different from that of patients who did not have liver cirrhosis. In addition, the median daily maintenance dose of voriconazole was significantly lower (2.13 mg/kg/d) than that of the control patients (6.27 mg/kg/d), suggesting that trough voriconazole concentrations are elevated in Child–Pugh class C patients. Thus, we conclude that oral voriconazole maintenance doses in patients with Child–Pugh class C should be reduced to approximately one-third that of patients with normal liver function, with the follow-up dose adjusted by therapeutic drug monitoring.