Step towards fault forecasting in hospital information systems

Nowadays, many organizations consider databases indispensable tools for their daily tasks. Particularly in healthcare units, databases have a vital role, since they archive very important information about patients´ clinical status, therefore, it is crucial that databases are available twenty-four hours a day, seven days per week. Healthcare units have already implemented fault tolerant systems, which intended to ensure the availability, reliability and disaster recovery of data. However, these mechanisms do not allow to take preventive actions in order to avoid the occurrence of faults. In this context, the necessity of the development of faults prevention and prediction systems is emerging. These systems can predict faults with some time in advance and provide taking early action to solve problems. The objectives of this paper are: monitor database performance and adapt a forecasting model used in medicine (Modified Early Warning Score - MEWS) to database reality.