Medical errors may result in loss of life and increase the costs of the healthcare system. Researchers has shown great potential of RFID (Radio Frequency Identification) technology in healthcare to prevent medical errors. However, existing international RFID standards do not include security specifications, which resulted in appearance of security threats. Actually, privacy issues are major obstacle to use this technology in hospital environments. This paper presents a novel approach of security for RFID systems based on a neural network, called tree parity machine. Moreover, we proposed a monitoring and audit system in the healthcare. This system is responsible for real-time patients monitoring and provide an audit in case of medical errors.