Inverse_transform on user input (Single input)

I have a dataset with just 4 categorical values
I have used LabelEncoder with OneHotEncoder with one field field43 as y variable or dependent
I get good accuracy of 0.67 with precision also … not bad.
I have train split and test as 0.25

Now when i provide a user input with this small amount of data, I get inverse transform as label not seen. Yes I have given onehotencoder handle unknown - ‘ignore’

Can i get pointers to examples where I can get inverse transform and get prediction? Guidance appreciated. To emphasize … I am looking for example on single user input which gives prediction. Thank you so much. Logan

Just a note: I am using SVM classifier