Predicting Low-Modulus Biocompatible Titanium Alloys Using Machine Learning

Area of Science:

• Materials Science
• Biocompatible Materials
• Computational Materials Science

Background:

• Titanium alloys are widely used in orthopedic and dental implants.
• There is a growing need for titanium alloys with a low Young's modulus and without cytotoxic elements.

Purpose of the Study:

• To analyze biocompatible titanium alloys using machine learning.
• To predict the composition of new titanium alloys with a low Young's modulus.

Main Methods:

• A database of 246 biocompatible titanium alloys was compiled, including composition and properties.
• Extra Tree Regression model was developed to predict Young's modulus.
• Monte Carlo simulations were performed to predict future alloy compositions.

Main Results:

• Specific heat was identified as the most influential parameter for lowering the Young's modulus.
• Machine learning models successfully predicted Young's modulus for titanium alloys.
• Simulations indicated the possibility of creating multicomponent alloys with a Young's modulus below 70 GPa.

Conclusions:

• Machine learning effectively analyzes and predicts properties of biocompatible titanium alloys.
• New titanium alloy compositions, primarily containing titanium, zirconium, tin, manganese, and niobium, can achieve desired low Young's modulus values.
• This research paves the way for developing advanced biocompatible materials for medical applications.