A hybrid framework of artificial intelligence-based neural network model (ANN) and central composite design (CCD) in quality by design formulation development of orodispersible moxifloxacin tablets: Physicochemical evaluation, compaction analysis, and its in-silico PBPK modeling

The objective was to apply the CCD-ANN system to design a QbD-based orodispersible tablet (ODT) of Moxifloxacin. The data sets of the trial formulations obtained from CCD were utilized for the training of the ANNbased model to obtain the optimized formulation. Three independent variables i.e. Acdisol, sodium bicarbonate, and compression force were selected to study their effect on critical dependent variables. The response variables were used for the supervised training using Holdback input randomization to develop a multi-layer perceptron (MLP) based ANN model for Moxifloxacin 400 mg ODT. The optimized formulation A generated by the prediction profiler was cross-validated by the CCD-based optimized formulation B using graphical and numerical methods. ANOVA findings revealed that there is no significant difference between the formulations A and B. These formulations were subjected to accelerated stability and shelf life was 31.380 and 25.475 months respectively. In-silico PBPK model exhibited comparative relative bioavailability of formulations A and B with the reference Moxifloxacin IR tablet. It can be concluded that ANN supported by CCD is a multi-objective simultaneous optimization technique, for pharmaceutical product development, especially when there is existing a nonlinear relationship between input and critical response variables.

Automated summary

The objective of this study was to design an orodispersible tablet (ODT) of Moxifloxacin based on QbD using the CCD-ANN system. The ANN-based model was trained using data sets obtained from CCD to obtain the optimized formulation. Three independent variables (Acdisol, sodium bicarbonate, and compression force) were chosen to study their effect on critical dependent variables. The optimized formulation A generated by the prediction profiler was cross-validated with the CCD-based optimized formulation B. ANOVA findings showed no significant difference between formulations A and B. Stability testing revealed shelf lives of 31.380 and 25.475 months for the two formulations respectively. The in-silico PBPK model showed comparable relative bioavailability of formulations A and B with the reference Moxifloxacin IR tablet.