Cardiovascular diseases are increasingly common in many countries. Cardiovascular risk (CVR) patients can continue their daily activities, but they must be monitored continuously in order to detect and respond to cardiovascular emergencies in the shortest possible time. Monitoring provides information to the medical personnel, which helps them make a diagnosis and decide upon a tailored treatment. This paper proposes a decision support system composed of three levels, which generates alarms when abnormal conditions occur in patients with CVR who are not hospitalized. The proposed system uses fuzzy methods of supervised and unsupervised classification to integrate qualitative representations of six physiological variables (heart rate, pulse strength, heart rate variability, temperature, movement and oxygen saturation). The first level determines whether the state is normal or alarm. The second level determines the degree of membership to sets of normal situations as well as to six possible groups of cardiovascular disease (Chest Pain, Syncope, Dyspnea, Metabolic Alterations Impacting Cardiovascular System or Miscellaneous). Finally, the third level determines specific membership to a selected group of cardiovascular emergencies. The proposed system proves to be useful for detecting abnormal patient events within the variables measured, and indicates degrees of membership to different emergency conditions in order to support the decision-making process of emergency responders.