*2.2. Coating Process and Design Parameters*

The use of ML algorithms is a promising approach [30] to not only qualitatively describe such interactions, which have to be determined in elaborate experiments, but also to quantify them [21]. Using ML, the aim is to generate reproducible, robust coating processes with appropriate, required coating properties. For this purpose, the main coating properties, such as coating thickness, roughness, adhesion, hardness and indentation modulus, of the coating parameter variations have to be analyzed and trained with suitable ML algorithms [31].

Within this contribution, the indentation modulus and the coating hardness are examined in more detail, since these parameters can be determined and reproduced with high accuracy and have a relatively high predictive value for the subsequent tribological behavior, such as the resistance to abrasive wear [32,33].

#### *2.3. Research Questions*

Resulting from the above-mentioned considerations it was found that existing solutions are solely based on a trial-and-error approach. ML was not considered in the specific coating design in joint replacements. So, in brief, this contribution wants to answer the following central questions. The first one is can ML algorithms predict resulting properties in amorphous carbon coatings? Based on this, the second one is how good is the resulting prediction of resulting properties in terms of quality and quantity? And lastly, can ML support in visualizing the coating properties results and the coating deposition parameters leading to those results? When ML can be used in these cases, the main advantages would be a more efficient approach to coating design with fewer to none trial-and-error steps and, lastly, the co-design of coating experts and ML. The following sections are to present the materials and methods used in trying to answer the stated research questions and provide an outlook on what would be possible via ML.

#### **3. Materials and Methods**

First, the studied materials and methods will be described briefly. In this context, the application of the amorphous carbon coating to the materials used (UHMWPE) as well as the setup and procedure of the experimental tests to determine the mechanical properties (hardness and elasticity) are described. Secondly, the pipeline for ML and the used methods are explained. Finally, the programming language Python and the deployed toolkits are described.
