Process Modelling Workshop, Bristol 2003 - page 1 Process and Conceptual Modelling of an Epidemiological Information System Tony Solomonides, Mohammed Odeh, Richard McClatchey (3) João Carriço, Jonas Almeida (2) Jean-Marie Le Goff (1) (1) ETT Division, CERN European Laboratory for Particle Physics, CH-1211 Genève 23, Switzerland (2) Instituto de Tecnologia Química e Biológica, Universidade Nova de Lisboa, Rua da Quinta Grande, nº6, 2780-156 Oeiras, Portugal (3) Faculty of Computing, Engineering and Mathematical Sciences, University of the West of England , Bristol BS16 1QY, United Kingdom The Role of an Epidemiological Information System A critical element in the strategy to decrease antibiotic resistance in pathogens is based on prevention and intervention at the level of medical practitioners and clinical care centres and is predicated on decreasing antibiotic consumption and implementing adequate infection control policies [1]. This can only be made possible if the bi-directional information flow between the molecular biology laboratories and clinical care centres is managed in a global, consistent, reliable, and timely manner and this necessitates the use of an information system. The proposed Epidemiological Information System (EIS) is a network of clinical care centres, molecular biology laboratories, and biomathematics or biometry units, sharing an information system to automate the integration of data acquisition, management and analysis of epidemiological data. The EIS presented in this paper was specifically designed to monitor and identify risk factors associated with carriage of antibiotic resistant bacteria in human populations. However, the proposed design may well be of wider applicability to other epidemiology surveillance programmes, e.g. bioterrorism and catastrophic animal diseases. The need to identify antibiotic resistant pathogens arises because there are already bacteria resistant to the majority or even all clinically relevant antimicrobial compounds and the pharmaceutical industry seems to have reached almost standstill in antimicrobial R&D. Indeed, several previously contained infectious diseases, such as multi-drug resistant tuberculosis, are now making a worrying comeback [1]. For the first time in 35 years, in April 2000, a completely new antibiotic compound (Linezolid) reached the pharmaceutical market [2], but a few months later, the first cases of microbial resistance to the drug were reported [3] [4]. Consequently, the control of infectious diseases relies increasingly on monitoring outbreaks and the timely intervention. Accordingly, the EIS proposed here was designed to enable a Bioinformatic Sentinel System intelligently to manage and respond to submitted clinical data within a framework of evolving epidemiological models. In addition to bringing together comprehensive data on patients and pathogens, the provision of an EIS enables the deployment of effective data mining tools. In extending conventional statistical analyses (multi- parametric statistical exploratory analysis), the large amount and diversity of data that need immediate incorporation in predictive models requires rigorous modelling through the use of artificial intelligence algorithms [5]. Those algorithms will continuously probe the constant influx of demographic, geographic and clinical data for risk factors related to the carriage and infection by antibiotic resistant bacteria. An additional critical feature of the proposed EIS is that it will be configured to send warnings/alarms to medical practitioners in the clinical care centres upon reaching a predefined threshold value for associations of risk factors to disease or carriage status. For example, a warning could advise on conducting further tests on one patient or quarantine for a cohort of patients, further hygiene measures in medical units or even a vaccination campaign.