Semantic Information Retrieval from Patient Summaries.

This paper presents a novel architectural model approach for the extraction of strictly necessary information from Patient Summary documents. These clinical documents, synthesizing the medical history of a patient, contain heterogeneous information, characterized by different confidentiality levels. For this reason, patients should be able to define privacy rules acting on the specific information they intend to share. However, the main system approaches currently used permit a patient to specify privacy policies only at document level. Thus, a patient is obliged to deny the access to the whole document to protect his/her privacy if only one kind of information is assumed particularly sensitive. To face this problem, the proposed architectural model is provided with semantic-enhanced features with the aim of allowing patients to define access rights based on document sections or information nuggets, trying moreover to fill the semantic gap between the queries and the information present in the documents.