Forecasting the state of pulmonary infection by the application of fuzzy cognitive maps

The presented research aims at the design and implementation of a medical decision support system for forecasting the state of pulmonary infection. As a part of such system, we propose a new method for modeling a decision making process. The model is learned using a population based algorithm in order to discover relationships among medical variables. We show how this model can be used to forecast the consequent state of the disease supporting this way the doctors' decisions on medical therapy. For the construction of the intended medical model we apply fuzzy cognitive maps (FCMs), an easily interpretable by physicians, graphical knowledge representation tool. The contribution of the paper is twofold. In the theoretical part, we propose a modified reasoning scheme for FCMs. In the experimental part, the developed model is validated in real patient data from Internal Care Unit (ICU) and the results of the performed simulations are outlined.