Automatically determining a root-cause indicator of a technical system concerning an output of a machine learning model

An Assistance apparatus for automatically determining a root-cause indicator of a technical system concerning an output of a machine learning model trained to analyze the technical system based on sensor data of different sensors measuring parameters of the technical system, comprising:

• at least one processor configured to;
• receive measured sensor data of the technical system and an output of the machine learning model for the measured sensor data as input, wherein the machine learning model is used to evaluate the technical system;
• determine an explanation information for the output of the machine learning model, comprising a set of feature values with one feature value indicating an impact of the feature on the output of the machine learning model;
• train a root-cause model based on root-cause labels assigned to the received output of the machine learning model to map the set of feature values of the explanation information onto a root-causes indicator;
• determine one of the root-cause labels by inputting a new set of feature values determined for an output of the machine learning resulting from inputting anew measured sensor data into the root-cause model;
• and
• output the determined root-cause label as root-cause indicator to a user interface wherein the root-cause indicator indicates a physical root-cause of the output of the machine learning model in the technical system and/or a measure to be applied onto the technical system and wherein the technical system is a device or machine in industrial manufacturing or heavy machinery.