I've never run across whatever it is that created the tokenizedDocument class shown in the table, but however the symptoms variable was created for the existing table, the same process must be used for your new T variable table content.
ADDENDUM: I see it (tokenizedDocument) is in the Text Analytics Toolbox.
Use it on the input variables of the new table, T
T=table('yourheadersstring',tokenizedDocument('yourdocumentfile'),'variableNames',{'headers','symptoms'});
dataset=[dataset;T];
Salt to match your variables...