Artificial Neural Network Prediction of Mechanical Properties in Mycelium-Based Biocomposites

Area of Science:

• Materials Science
• Biotechnology
• Artificial Intelligence

Background:

• Mycelium-based biocomposites (MBBs) offer a sustainable alternative to synthetic materials.
• Optimizing MBB mechanical properties is challenging due to complex interactions between substrate, fungal species, and processing.

Purpose of the Study:

• To develop and validate an artificial neural network (ANN) model for predicting MBB mechanical properties.
• To assess the predictive accuracy of the ANN for internal bonding (IB) and compressive strength (CS).

Main Methods:

• An ANN model was trained using experimental data on substrate composition, fungal species, and physical properties.
• ANN predictions for IB and CS were compared against measured values.
• Scanning electron microscopy (SEM) was used to visualize microstructural heterogeneities.

Main Results:

• The ANN demonstrated high predictive accuracy, with R² values of 0.992 for IB and 0.979 for CS.
• Internal bonding (IB) predictions were more precise than compressive strength (CS) predictions.
• Specific fungal species (Ganoderma sessile, Trametes versicolor) showed high IB; Trametes versicolor's CS varied significantly with substrate quality (virgin vs. recycled wood).

Conclusions:

• ANNs are effective tools for predicting MBB mechanical properties, significantly reducing the need for extensive experimental testing.
• Substrate quality and fungal species interactions critically influence MBB performance.
• This AI-driven approach accelerates the development and characterization of sustainable biocomposites.