Smart healthcare system in dietary behavior recommendations based on physiological data

Theoretical basis and empirically-derived model for integrating body sensor data for personal health promotion through lifestyle recommendations are lack. This paper develops and evaluates a smart healthcare system serving as a decision support aid for health professionals in support of patients with metabolic syndrome. A reasoning algorithm in the system is proposed to generate suggested recommendations in dietary behavior based on each patient's specific physiological conditions, in which a small number of dietary behavior advice points are initially provided as the input of the algorithm. To evaluate the system, the system-generated recommendations are compared with the recommendations manually modified by medical specialists. The system accounts for the five physiological indicators used for metabolic syndrome diagnosis, producing 134 distinct clusters of recommendations accounting for each combinatorial risk level groups, using ten manually-derived clusters of dietary behavior points as the input. The comparison results indicate average compliance of 71.53% for dietary behaviors advice. This indicates the system has the potential to support health professionals in the process of providing personalized advice. The system would be particularly useful in situations where the use of increased physiological data increases the quantitative effort required to produce such recommendations.