Machine Learning Assisted Optimization Of Titanium Dioxide Nanotube Micropattern And Its High-throughput Applications

The bipolar electrochemical reaction devices are simple and easy to operate,and are widely used to fabricate micropattems with structure/property gradients.However,bipolar electrochemical reactions involve many experimental parameters,and the relationships between experimental parameters and micropatterns’ structures and properties are intricate.Hence,it is challenging to optimize experimental parameters and obtain the ideal micropattern efficiently and quickly.Given the current key technology and important scientific issues,including the construction of gradient TiO2 nanotubes micropatterns(TNMs)on biomedical titanium surface and their applications in high throughput study,this thesis focuses on optimizing TNMs using machine learning methodologies and applying TNMs for high-throughput studies in biomedical science.The main research contents are as follow:(1)Optimization of TNMs using active learning,including TNMs preparation,classification,regression,and verification.Given sometimes titanium foils would rupture from the middle and fail the experiment,leaving no measurable data,we utilize a binary classification model to convert it into a binary classification.The decision tree model distinguishes the fused sample from the standard samples,defines the experimental boundaries,and analyzes the importance of experimental factors.(2)The gradient boosted regression tree(GBRT)analyzes the valid data screened from the classification model and provides predictions and directions for optimizing TNMs.Then,predictions are verified by experiments,and new results update the training dataset for the next learning loop.Within several iterations,we obtain the optimal TNMs with a diameter range of 27-470 nm,expanding the gradient to the largest extend without tedious experiments.(3)Using electrochemical deposition,we construct micropatterned silver nanoparticles on optimized gradient TiO2 nanotubes(TNMs-AgNPs),with morphology and size gradient distribution.The results show that the size and number of Ag nanoparticles(AgNPs)increase with the increment of nanotube diameter.(4)TNMs and TNMs-AgNPs were used for a high-throughput study of cell adhesion/proliferation and bacterial adhesion.Results show that adherent cell numbers and cell spreading areas decrease with the increment of nanotube diameter,regardless of the presence of silver nanoparticles on the TNMs’ surface.For the nanotube surfaces with the same size,cell numbers and cell spreading areas reduce on the nanotubes with AgNPs,indicating AgNPs restrain cell activity.Bacterial adhesion experiments show that for TNMs without AgNPs,more bacteria adhere to small-sized nanotubes with no significant difference on other nanotubes.However,with the increment of nanotube diameter,adherent bacteria numbers diminish on the AgNPs doped TNMs,indicating that the deposited AgNPs had a significant inhibitory effect on bacterial activity.