Key points are not available for this paper at this time.
At present, the elderly population commonly faces difficulties in accessing medical services, and traditional human companion models fail to meet the rapidly growing demand for medical assistance. This paper puts forward a companion robot for the elderly that integrates the random forest algorithm with the Kano model of customer satisfaction. By identifying user satisfaction levels with different functions, the robot automatically maps user profile characteristics to corresponding function preferences. Digital twin modeling technology records users’ long-term behavior and provides feedback during voice interactions with elderly users. Performance validation is conducted using a self-constructed dataset. The results show that the root mean square error in user needs analysis averages 0.173, with a median of 0.168. After 1000 iterations, the accuracy stabilizes at 97.2%, and the F1 score reaches 0.97. In a simulation experiment, the robot achieves a recognition precision of 96.8% and a response time of 0.85 s, outperforming the comparison robot with 1.12 s. It also maintains a power consumption of 44 W and a behavior adaptation rate of 91%. These findings demonstrate that the proposed robot performs reliably and excels in emotional companionship, offering new possibilities for applying intelligent robots in the medical service field.
Wen et al. (Mon,) studied this question.