MENELAS

An Access System for Medical Records using Natural Language

Contact:
P. Zweigenbaum
Service d'Informatique Médicale, Assistance Publique - Hôpitaux de Paris
91, boulevard de l'hôpital, F-75634 Paris Cedex 13, France
email: pz@biomath.jussieu.fr
Source Reference:
P. Zweigenbaum, Consortium Menelas. MENELAS: An Access System for Medical Records Using Natural Language. Computer Methods and Programs in Biomedicine, 45:117--120, 1994.

The overall goal of project Menelas is to permit the exploitation of the medical information held in patient discharge summaries written as free text. The realisation of the pilot system is based on natural language processing and information retrieval techniques.

Author: Jacques Bouaud (jb@biomath.jussieu.fr)

Modern medicine uses an increasing quantity of information. This is due to the emergence of new, diverse and complementary investigation techniques (imaging, etc.), and mostly to growing numbers of patients treated for chronic diseases in the developed countries. The double goal of enhancing the quality of health care and of controlling costs implies a need for rapid access to information available on patient history. This creates an increasing demand for better management of medical information resources, and of patient records. In the case of pathologies that are sufficiently limited and well standardised in their treatment, computerised medical records, developed in the framework of Hospital Information Systems, are being used. However, they require users to adhere strictly to a fixed, closed set of items to describe patient information.

Nevertheless, a number of applications such as ongoing patient care, evaluation of health care and clinical research use information which is predominantly available in narrative patient discharge summaries, in the form of "free text." These reports synthesise the relevant information concerning each patient, and contain a wealth of data regarding patient history, current diagnoses, and current and suggested subsequent treatments. They constitute a widely available, low cost source of reliable information, which is flexible and can adapt to medical progress. However, available industrial technology does not allow access to this information. Project Menelas aims to design and implement a new sort of computer system using natural language processing and information retrieval techniques to fill this gap. The pilot system will offer the following services:

Faced with an atypical patient, clinicians can search the stored reports for patients with similar characteristics, and find the treatments they received [diagnosis aid].
The manager of a hospital department may consult an electronic activity board fed with data extracted from the reports [self-auditing, medico-economic management].
Nomenclature codes are automatically produced for each report, and can be included in standardised discharge summaries [medico-administrative management].
Public health researchers can test clinical research hypotheses on the basis of the available patient sample [clinical research and public health].

Menelas is a multilingual project involving French, English and Dutch. The system employs techniques from the Artificial Intelligence and Computational Linguistics communities that draw on linguistic, commonsense and medical domain knowledge sources in order to transform ordinary texts into a calculable form. Menelas will not only enable access by contents to patient discharge summaries, but also, when appropriate, produce a standardised representation of extracted information, in the form of international nomenclature codes: ICD-9-CM codes (International Classification of Diseases). The system is designed with a concern for extensibility to different application domains; it is tested in the domain of coronary diseases. Started in January 1992, the project will end in December 1994.