ADAPTIVE PREDICTION OF CHANGES OF PHYSIOLOGICAL/PATHOLOGICAL STATES USING PROCESSING OF BIOMEDICAL SIGNALS

A method and system predicts changes of physiological/pathological states in a patient, based on sampling, processing and analyzing a plurality of aggregated noisy biomedical signals. A reference database of raw data streams or features is generated by aggregating one or more raw data streams. The features are derived from the raw data streams and represent physiological/pathological states. Each feature consists of biomedical signals of a plurality of patients, wherein several patients have one or more of the physiological/pathological states. A path, which is an individual dynamics, between physiological/pathological states is obtained according to their order of appearance. Then, a prediction of being in physiological/pathological states, or transitions to physiological/pathological states in the patient, is obtained by comparing the individual dynamics with known dynamics, obtained from prior knowledge.