Automated classification of free-text pathology reports for registration of incident cases of cancer.

OBJECTIVE: Our study aimed to construct and evaluate functions called "classifiers", produced by supervised machine learning techniques, in order to categorize automatically pathology reports using solely their content. METHODS: Patients from the Poitou-Charentes Cancer Registry having at least one pathology report and a single non-metastatic invasive neoplasm were included. A descriptor weighting function accounting for the distribution of terms among targeted classes was developed and compared to classic methods based on inverse document frequencies. The classification was performed with support vector machine (SVM) and Naive Bayes classifiers. Two levels of granularity were tested for both the topographical and the morphological axes of the ICD-O3 code. The ability to correctly attribute a precise ICD-O3 code and the ability to attribute the broad category defined by the International Agency for Research on Cancer (IARC) for the multiple primary cancer registration rules were evaluated using F1-measures. RESULTS: 5121 pathology reports produced by 35 pathologists were selected. The best performance was achieved by our class-weighted descriptor, associated with a SVM classifier. Using this method, the pathology reports were properly classified in the IARC categories with F1-measures of 0.967 for both topography and morphology. The ICD-O3 code attribution had lower performance with a 0.715 F1-measure for topography and 0.854 for morphology. CONCLUSION: These results suggest that free-text pathology reports could be useful as a data source for automated systems in order to identify and notify new cases of cancer. Future work is needed to evaluate the improvement in performance obtained from the use of natural language processing, including the case of multiple tumor description and possible incorporation of other medical documents such as surgical reports.

An objective method for bed capacity planning in a hospital department - a comparison with target ratio methods.

OBJECTIVES: To propose an objective approach in order to determine the number of beds required for a hospital department by considering how recruitment fluctuates over time. To compare this approach with classical bed capacity planning techniques. METHODS: A simulated data-based evaluation of the impact that the variability in hospital department activity produces upon the performance of methods used for determining the number of beds required. The evaluation criteria included productive efficiency measured by the bed occupancy rate, accessibility measured by the transfer rate of patients due to lack of available beds and a proxy of clinical effectiveness, by the proportion of days during which there is no possibility for unscheduled admission. RESULTS: When the variability of the number of daily patients increases, the Target Occupancy Rate favors productive efficiency at the expense of accessibility and proxy clinical effectiveness. On the contrary, when the variability of the department activity is marginal, the Target Activity Rate penalizes the proxy of clinical effectiveness, and the Target Occupancy Rate underoptimizes productive efficiency. The method we propose led to a superior performance in terms of accessibility and proxy of clinical effectiveness at the expense of productive efficiency. Such a situation is suitable for intensive care units. In the case of other departments, a weighting procedure should be used to improve productive efficiency. CONCLUSIONS: This approach could be considered as the first step of a family of methods for quantitative healthcare planning.

Monitoring incident report in the healthcare process to improve quality in hospitals.

Healthcare organizations are facing growing pressures to adopt intelligent technology to promote quality and safety care in public and private hospitals. In 2000, the Institute of Medicine report also indicated that an estimated 44,000 to 98,000 Americans die annually as a result of preventable medical errors and it appears that information management in hospitals can help the organization to improve the quality level. This paper aims to present an experience in the management of events not compliant with the best practice by monitoring these events in a hospital. We used ISO standard to implement general quality process and quality management. This project consists in proposing the possibility of declaring different dysfunctions and incidents by a simple form integrated into the intranet services of the hospital for the medical, nursing and administrative staff. This should lead to quality management of the medical units.

Managing educational resource in medicine: system design and integration.

The objective of this paper is to describe our experience in developing a tool based on web technologies, for storing, managing and providing medical courses written by professors in the medical university of Rennes. The increasing number of documents sent by professors leaded us to build a specific resource management system. We created a relational database, containing all meta information about each available document. Professors provide their courses in various formats. We use natural language parsing techniques to extract information from the text, and provide a proper semantic indexation which will be used by a medical-specific search engine. Then the content of our database is dynamically displayed on a web interface. A user's directory identifies teachers and students, controls the access, tracks the students' navigation and allows an on-line discussion forum. This portal contains 524 courses and we had more than 3,000,000 connections on it last year. We are now integrating its content using the semantic web approach in a larger project: the French Virtual Medical University.

A simple method to optimize hospital beds capacity.

OBJECTIVE: The number of acute hospital beds is determined by health authorities using methods based on ratios and/or target bed occupancy rates. These methods fail to consider the variability in hospitalization demands over time. On the other hand, the implementation of sophisticated models requires the decision concerning the number of beds to be made by an expert. Our aim is to develop a new method that is as simple to use as the ratio method while minimizing the roundabout approaches of these methods. METHOD: A score was constructed with three parameters: number of transfers due to lack of space, number of days with no possibility for S unscheduled admissions and number of days with at least a threshold of U unoccupied beds. The optimal number of beds is the number for which both the mean and the standard deviation of the score reach their minimum. We applied this method to two internal medicine departments and one urological surgery department and we compared the solutions proposed by this method with those put forward by the ratio method. RESULTS: The solutions proposed by this method were intermediate to those calculated by the local and national length of Stays ratio methods. Simulating an unusual increase in admission requests had no consequence on the bed number selected, indicating that the method was robust. CONCLUSION: Our tool represents a real alternative to the ratio methods. A software has been developed and is now available for use.

VUMeF: extending the French involvement in the UMLS Metathesaurus.

A considerable number of robust vocabularies and thesauri have been developed for the healthcare and biomedical domain. No single vocabulary, however; provides complete coverage of the information needs from a public health perspective. The results of an investigation of vocabulary sources for the development of a comprehensive controlled vocabulary for the public health domain at the Centers for Disease Control and Prevention (CDC) is presented.

A preliminary study to assess a French code of ethics for health teaching resources on the Internet.

BACKGROUND: Constant assessment of the quality of health information on the Internet is an absolute necessity as peer review is often lacking in this media. OBJECTIVE: To develop a simple and easy French Code of Ethics, which will enable medical students to judge quality of health information in teaching material available on the Internet. DESIGN: Three medical informaticians selected ten criteria from previously established codes of ethics from Europe and the USA. This instrument was tested on a sample of 30 health Internet teaching resources. RESULTS: For the panel of experts, chance corrected inter-observer agreement (kappa) for quality rating ranged from k = -0.19 and k = 0.33, demonstrating poor agreement among the raters. CONCLUSION: If negative results of this preliminary study are confirmed by further research, this finding may detrimentally affect projects in Europe to accredit or certify Internet health resources.

DIAMED: a probabilistic diagnostic aid system on the web.

DIAMED is a system to assist the physicians in the diagnostic process using probabilistic networks as knowledge representation. These networks make it possible to reason on medical data by applying Bayesian methods and to take into account uncertainties of the facts in the resolution of the clinical cases. The proposed model re-uses knowledge contained in an existing knowledge base (ADM). An interface of DIAMED developed on a Web server remotely assists the experts of each medical specialty in updating and validating the knowledge base. Most of the data processing is automated while being based on information preexistent in the ADM base : Constitution of lexicons starting from the existing dictionaries of the ADM system, are then used to work out the requests for selection and update of the knowledge base. One of its assets resides in its pseudo-segmented structure in several layers. The propagation of information is thus limited to only one part of the probabilistic network and calculations are therefore limited.

A cognitive approach for designing a computer-based learning tool in dentistry: ODONTOS application.

This work emphasises a cognitive approach to build ODONTOS, a computer-based education system in dentistry. It aims at providing for both instructors and students without any particular knowledge in computer programming; a tool dedicated to dental disease learning. For instructors, it implements for knowledge construction e.g. creating and proposing clinical cases according to student learning style, using multimedia technology. For students it is a tool, which privileges their learning style in order to solve uniform and inflexible learning problem. World Wide Web (WWW) technology has allowed us to offer flexible learning program without any pressure of time and place to student. 13 instructors and 84 students tested the ODONTOS prototype via Internet and answered to the questions related to their habit regarding learning style for the students and knowledge construction for the instructors. Although video data have emerged as an important learning support for the students (82 out of 84), their use in knowledge construction still remains unusual comparative with the use of text and picture (10 out of 13 instructors). It appears that, the integration of video data should be more used by instructors (2 out of 13) while textual data are overwhelmingly used in knowledge construction (13 out of 13). We introduce ODONTOS, and discuss the concrete contributions of this work.

Design of a CSCL environment for clinical reasoning learning and problem-based learning in medicine.

The medical curriculum is changing, student-centered learning is currently used in medical schools. Problem-based Learning and Clinical Reasoning Learning develop the students' reasoning strategies. CSCW (Computer-Supported Cooperative Work) technology is used in Problem-Based Learning systems. We have designed a CSCL (Computer-Supported Collaborative Learning) environment for improving group coordination and communication in Clinical Reasoning Learning sessions. To support these new educative technologies, a prototype has been developed.

Sharing medical data for patient path analysis with data mining method.

The Agora Data project started in October 1997 in France. The objective was to share medical data between several medical institutions to analysis medical care pathways for patients that suffer from low back pain. The analysis of the medical records decomposed in three steps allowed us to produce knowledge on medical contacts of patients with the health care system. In order to study the relations between these contacts, we created medical path of patients within the framework of the possible contacts we had isolated. This work relates the implementation and the first results of the pilot study.

[An experience in teaching pharmacology on the internet]. / Une expérience d'enseignement de la pharmacologie par Internet.

Internet and Intranet are omnipresent in the University world. We developed an easy-access website (www.med.univ-rennes1.fr/etud/pharmaco) devoted to teaching pharmacology and report here our experience after 4 years of use. Our objective was to determine the value of this new teaching tool in the medical coursus. The site is entirely free and presents approximately 50 topics and diaporamas discussing various themes: the cell, the receptor, general pharmacology, clinical research, population research, drug classes, etc. Harbored by the Medical Informatics Laboratory of the University of Rennes Medical School, this site serves as a reference for medical students and others. More than 100 visits are recorded daily, approximately half from visitors outside France. The advantages of this new teaching tool, which operates within the framework of a Virtual Medical University project, are evident for students and professors alike. Its impact on the quality of drug therapy by future doctors remains to be determined.

Issues in the design of medical ontologies used for knowledge sharing.

Recent work in Medical Informatics is exploring the development and the use of formal ontologies as a way of specifying content-specific agreements for the sharing and reuse of knowledge among several computer systems. We describe the role of ontologies in supporting knowledge sharing activities in medicine Principles for the design of ontologies have been proposed, mainly in other domains: these principles include parsimony, clarity, representation of categories versus terms, and coherence. We analyze how and why these principles can or cannot be applied from case studies from medical systems. Regarding the fact that most of medical concepts are empirical, selected design decisions are discussed. An alternative representation choice consists in mapping principled general core ontologies and domain ontologies.

Knowledge acquisition to qualify Unified Medical Language System interconceptual relationships.

Adding automatically relations between concepts from a database to a knowledge base such as the Unified Medical Language System can be very useful to increase the consistency of the latter one. But the transfer of qualified relationships is more interesting. The most important interest of these new acquisitions is that the UMLS became more compliant and medically pertinent to be used in different medical applications. This paper describes the possibility to inherit automatically medical inter-conceptual relationships qualifiers from a disease description included into a database and to integrate them into the UMLS knowledge base. The paper focuses on the transmission of knowledge from a French medical database to an English one.

HIV/AIDS epidemic in French Guiana: 1979-1997. Groupe d'Etude Clinique de l'Infection VIH en Guyane Française.

The incidence of AIDS in French Guiana remains one of the highest in Latin America and the Caribbean. The annual AIDS incidence rate increased continually from the start of the epidemic until 1995, when it reached 59.3/100,000 population declining thereafter to 26.6 in 1997. The prevalence of HIV in pregnant women was 0.9% in 1993, increasing to 1.3% in 1995, and that in individuals attending sexually transmitted disease (STD) clinics was 2.1% in 1996. We included 224 patients in a study of survival after AIDS diagnosis. The principal AIDS-defining diagnosis was tuberculosis in 20.5% of reported cases. The median duration of survival was 10.2 months. Multivariate analysis showed that, patients > or = 45 years at entry progressed more rapidly to AIDS than younger patients. HIV prevention and access to health care should be developed in the various ethnic communities and adapted to cultural status. The progressive implementation of multiple antiretroviral therapies since 1996 may further reduce progression of the disease but early HIV diagnosis is required to improve the overall prognosis of HIV-infected patients.

Radiology on Internet: advice in consulting websites and evaluating their quality.

Services offered by Internet are increasing continually and 15,000 medical websites covering all the specialties are available presently. Finding relevant information with a spider-web organization is difficult. We recommend starting with hierarchical lists which propose a selection of sites corresponding to medical specialties. We give our selection of important radiological websites dealing with image databases, case reports, radioanatomy and continuing medical education. While surfing the Web, an evaluation of the quality of websites is necessary. We used quality criteria proposed by Darmoni to rate the quality of ten websites. The global quality is good; help pages and external links are the main noticed shortcomings. All webmasters should pay attention to quality criteria and show visible marks of Darmoni criteria on their home page. To be able to foresee which Internet options should be developed, we evaluated our website and sent a questionnaire to our users. Clinical cases are the most requested facilities; they could be used for continuing medical education.

[Web radiology servers: advices for users, their development and evaluation of the quality of sites]. / Les serveurs web de radiologie: conseils pour leur consultation, leur réalisation et mesure de la qualité des sites.

Sites available on the Internet are constantly increasing and over 15,000 medical websites covering all medical sub-specialties are now available. Finding relevant information on the web may be time consuming and difficult. We suggest that new users begin by accessing servers where lists of available sites are grouped under headings such as medical sub-specialties. We will list a selection of useful radiological websites providing image databases, case reports, radiological anatomy and continuing medical education. While surfing on the web, an evaluation of the quality of the websites is necessary. We used the quality criteria proposed by Darmoni to rate the quality of ten websites. The overall quality was good, but the lack of help options and external links were the main pitfalls that we noted. All webmasters should pay attention to quality criteria and show the Darmoni criteria on their homepage. In order to evaluate which internet options should be developed, we evaluated our own website and submitted a questionnaire to our users. Clinical cases were the most requested feature, they could be used for continuing medical education.

A computer-assisted drug prescription system: the model and its implementation in the ATM knowledge base.

Informatisation of drug prescription is an important topic in medical informatics. For several years now, computerized drug databases have been implemented. Usually only a small part of the prescriptions can be stored in prescription systems because of the format of the included information; prescriptions contain essentially free text without any structure and homogeneity of the used vocabulary. In this article a model is presented for knowledge representation in a computerized drug prescription system. The model should be applicable to clinical practice and be didactic for medical students. The problem of standardization of terminology had to be solved. A computer-assisted drug prescription program has been developed. The next step is its validation by clinicians. The program can also be used in a consultation mode.

CARDIOMEDIA: a communicable multimedia medical record on Intranet and digital optical memory card.

As part of French health reform, French physicians were computerised by the end of 1998. A specific Intranet network will be used to communicate medical data between the health professionals. The objectives of the CARDIOMEDIA project were to develop and evaluate the feasibility of a coronary multimedia data record stored on an optical card and communicable on Intranet within the hospital. Patients treated by angioplasty at the University Hospital of Rennes participated in the experiment. In general, patients are treated in the University Hospital and are followed up by another health care provider closer to their home. The patient leaves the University Hospital with his card, which is directly available elsewhere for emergency or for consultation. This approach is assumed to reduce the number of examinations and to offer a better patient follow-up. The CARDIOMEDIA card is a specialised record including various data types: text, images, image sequences of coronarography and ECG signals. For this purpose an optical card with its large memory is very convenient. We used the DICOM format for image exchange and management. It is combined with CARDIOMEDIA specific compressing software. For the multimedia record the HTML format and web Intranet method are chosen. This provides an intuitive interface which can combine various data types and helper applications like a DICOM image viewer.

A method for an automated decomposition of a complex term into its conceptual components.

The paper deals with the improvement of the MAOUSSC model (Modèle d'Aide et d'Orientation d'un Utilisateur au Sein des Systèmes de Codage) and system. Its specific purpose is the automation of the description of medical and surgical procedures. We have developed an automatic decomposition method using a linguistic and conceptual approach based on the UMLS knowledge base. This work concerns the processing of 100 procedure wordings from the digestive surgery domain. We introduce a prototype of such a system automating the decomposition through a web interface.