Converting a data frame (tabular data) into a graph to use GNN for multi-task learning?

I have a data frame as follows:

    mydata_frame:
                 IDs     f1        f2      f3  ...  Target1  Target2  Target3
                 Id1     5.3       6       21  ...    1        0         1
                 Id1     0.4       0       0   ...    1        0         1   
                 Id1     21.2      1       4   ...    1        0         1
                 Id1     16        5       17  ...    1        0         1
                 Id1     30        2       5   ...    1        0         1
                 Id1     9         1       1   ...    1        0         1
                 ...     ...        ..      ..      .. 
                 Id2     8         9       10  ...    0        1         1
                 Id2     0         0       0   ...    0        1         1                 
                 Id2     12        10      7   ...    0        1         1
                 Id2     4         1       3   ...    0        1         1
                 Id2     22        25      3   ...    0        1         1
                 Id2     10        15      4   ...    0        1         1

In this data, each specific ID (e.g., Id1) has three targets variables expressing three different tasks. I would like to formulate this task as a multitask learning and use GNN to train and predict the different tasks. I was wondering how can I convert or represent this data as a graph so I can use GNN to train multiple tasks.