To achieve this, we start by

1. splitting the text by newline character ('\n') split(variables('Source'),outputs('New_Line'))
2. store headers for future reference outputs('Split_text_lines')?[0]
3. filter array to find lines that contain 'EMPTY' contains(item(),'EMPTY')
4. join the lines back with headers by newline character ('\n') join(union(outputs('Store_Headers'),body('Filter_lines_that_contain_EMPTY')),outputs('New_Line'))

We preserve headers so that final text files does not get ambiguous data.

P.S. power automate does not treat '\n' string well so we achieve this by defining a compose and placing a new line there as follows (hit enter in input section to achieve this)

Here's the full implementation