Past Medical History (PMH)

Past Medical History (PMH) sections of a large corpus of clinical notes, as a proxy for problem lists. Findings show that when listing patients’ history, physicians convey several semantic types of information, not only problems. Furthermore, they often group related concepts in a single line of the PMH. In contrast, traditional problem lists allow for only a simple enumeration of coded terms.

The past medical history (PMH), including past surgical history, stood out as the only longitudinal summary of patient status. he PMH tends to be a list of relevant medical data on the patient. Its structure is not dissimilar to that of a problem list, but more complex.

The PMH is entered in a free text field, not in a structured format. To diagnose a patient, a physician must first develop a comprehensive understanding of the patient’s medical status, including preexisting problems.

Today’s patient record may contain an overwhelming amount of information, and physicians can struggle to identify all salient information, especially when pressed for time for effective diagnosis.

Methods

Clinical Document Collection

A collection of 7673 initial visit notes (corpus) was obtained from the Columbia University Medical Center Milstein Hospitalist Service.

All notes from the Service are entered through semi-structured entry templates in a system called eNote

The notes were stored using the Clinical Document Architecture (CDA) XML schema. This allowed for a simple XSL transformation (XSLT) to filter protected health information (PHI) and convert sections of interest to text.

A small Java application was written to perform this XSLT on each document and do basic preprocessing to prepare the text for natural language processing analysis.

Data Preparation

The corpus was then parsed with the MedLEE natural language processor to obtain the semantic structure and UMLS codes of concepts represented in these notes.