Investigation of hydroxyapatite dicalcium phosphate scaffold properties using a Lamarckian immune neural network

In this paper, a hybrid predictive tool is developed to model the main properties of Hydroxyapatite Dicalcium Phosphate Anhydrous (HA/DCPA) scaffold. Since this biodegradable scaffold is widely used as a bone substitute material, it is mandatory to develop an identification model which can give some useful information about its pivotal characteristics. To this aim, two main steps are made. Firstly, the effect of HA/DCPA weight ratio on its compression strength, elastic modulus, calcium dilution, density, porosity and weight change are investigated through experiments. Thereafter, the proposed tool which integrates an Aggregated Neural Network (ANN) and a well-known optimisation method called Multi-objective Lamarckian Immune Algorithm (MLIA) is used to find a robust model. To elaborate on the effectiveness of the proposed approach, some well-known identification systems such as Back Propagation Neural Network (BPNN), Adaptive Neuro-Fuzzy Inference System (ANFIS) and Composite Neuro Particle Swarm Algorithm (CNPSA) are used. The results indicate that the method is strongly capable of modelling all the properties of HA/DCPA simultaneously.