With the increasing percentage of elderly population in modern society, more attention is paid to improve the living quality for elderly who live alone at home. However, most home care units were designed to tackle single problem, not omni and real-time systems, faced at home. This study proposes an intelligent approach to identify elderly body information. Two sub-modules, fingertip temperature measurement and intelligent cane, are devised in this study. We use infrared temperature sensors to design a fingertip temperature measuring module that tracks and predicts the elderly fingertip temperature variations. In the case when the predicted value deviates from the ambient temperature to a present amount, an alarm is automatically sent to the contact persons or caregivers for taking care of the abnormal situations. Parkinson's disease patients suffer from muscular rigidity, slowness of the movement, and irregular trembling. They may take along a cane when go out for a walk. Tri-axial accelerometers are embedded into a cane to continue monitoring the patient's gaits as well as to detect if falling occurs. Based on the detected signals from accelerometers a fuzzy inference model is proposed to identify the moving patterns on the patient. Experimental results are presented to verify the effectiveness of the proposed system.