De-identification causes loss of non-PHI information in discharge summaries

[abhinayvyas]

Hi all,

Thanks for creating this helpful dataset. While looking at physician notes, I have noticed loss of important non-PHI information. For patient 10250007 under hadm id 25898366, I have noticed deidentification for parameters that explains completeness of health check. Under physicial examination section, "Glasgow" for Glasgow Coma Score is de-identified and segments for certain questions have been masked as well since they are formatted from multiple choice questions.