Personalized drug concentration predictions with machine learning: an exploratory study

Background: The dose individualization by therapeutic drug monitoring (TDM) can be improved if population-based reference ranges are available, as there is large inter- and intrapatient variability. If these ranges are not available, dose individualization may not be optimal. Machine learning can help achieve accurate drug dose settings and predict the resultant levels.Methods: Two random forest models, a multi-class classifier to predict dose and a regression model to predict blood drug level were trained on 320 patients’ data, consisting of their age, sex, dose and blood drug level. The classifier consisted of 1000 estimators (decision trees) and the regression model consisted of 1300 estimators. The model was evaluated on randomly split test set having 10% of the total dataset size. The regression model was compared against k-Nearest neighbor and linear regression models. The classifier was evaluated using accuracy, precision, and F1 Score; the regression model was evaluated using R2, Root mean squared error, and mean absolute error.Results: The classifier had an out-of-sample accuracy of 68.75%, average precision of 0.7567, and an average F1 score of 0.6907. The regression model had an out-of-sample R2 value of 0.2183, root mean squared value of 3.7359, and a mean absolute error of 2.5156. These values signify an average classification performance, and a below-average regression performance due to small dataset.Conclusions: It is possible for machine learning algorithms to be used in therapeutic drug monitoring. With a well-structured, rich, and large dataset, a very accurate model can be built.

Effects of Broad Spectrum Antibiotics on Measurement of Immunosuppressant Drugs

Background: Antibiotics used parenterally can affect blood drug level measurements, as measured in diagnostic tests. Objective: To investigate the effect of six different antibiotics commonly used in intensive care units on tacrolimus, sirolimus, everolimus and cyclosporin A levels measured by mass spectrometry. Methods: Ampicillin + sulbactam (AB1, IV, 1 g), imipenem + cilastatin sodium (AB2, IV, 500 mg), piperacillin + tazobactam (AB3, 4.5 g, IV), ertapenem (AB4, IV, 1 g), meropenem trihydrate (AB5, 500 mg, IV) and ceftriaxone (AB6, 1 g, IV) antibiotics were used for the interference assay. Measurements were performed on the Shimadzu 8045 (Japan) LC-MS/MS instrument. Bias values were calculated. Results: The least affected immunosuppressant was cyclosporine A (between -6.88% and 3.40%). The most affected were everolimus and sirolimus. Ertapenem caused negative interference on the level of everolimus at the rate of -27.34% and sirolimus at the rate of -26.79%. Piperacillin + tazobactam and imipenem + cilastatin sodium caused positive interferences on sirolimus at the rate of 24.24% and 22.73%, respectively. Ampicillin + sulbactam, meropenem trihydrate and ceftriaxone affected the sirolimus levels at lower rates (-4.49%, 5.93% and 9.86%). Everolimus levels deviated at the rate of -11.21% to -16.99% due to imipenem + cilastatin sodium, meropenem trihydrate and ceftriaxone. Conclusion: This study demonstrated the potential of antibiotic use affecting immunosuppressant levels. Antibiotic interference, especially in transplant patients, may cause erroneous immunosuppression, increasing the likelihood of rejection.

Anti-epileptic drug intake adherence: the value of the blood drug level measurement and the clinical approach

It was evaluated the patient antiepileptic drug (AED) intake adherence in a pilot cross-sectional study carried out at a neurologic out-patient clinic of a university hospital. Ninety-three AED blood concentration (phenobarbital, phenytoin, carbamazepine) were analyzed from 24 patients. The variability of the AED blood level was measured (in the steady state period by means of the variation coefficient) and compared with the self-reported antiepileptic medication non-adherence, AED blood level according to the range (therapeutic or not), and the seizure control. It was not observed any strong correlation between the higher value of variability and the other three parameters of no adherence. The highest correlation was with the blood drug level (therapeutic or not). The evaluation of blood drug measurement alone, except in cases of extreme low adherence and variability of drug intake, is not enough for the recognition of incorrect drug intake, but the clinical markers and the self-reported adherence have to be also considered for this sort of evaluation.

Imipramine and Amitriptyline Plasma Concentrations and Clinical Response in Major Depression

Plasma drug concentrations and clinical response were measured in two groups of hospitalised depressed patients, who received amitriptyline or Imipramine double-blind, in a dosage of 250 mg for four weeks. Virtually no significant linear or curvilinear relationships were found between any plasma measure and any measure of clinical response. Modest but significant direct relationships were found between age and concentration of parent drugs but not demethylated metabolites. Blood drug level measurement therefore appears to be of little value in monitoring drug treatment of depressed in-patients.