This paper discussed the concept of self-sustained treadmill using physical activity of humans. During the step-by-step walking on treadmill calories burned and using this concept can power up the treadmill.  It also incorporated the power optimization using installation of PZT tiles on treadmill and prototype model also incorporated in the study. The deployment of PZT-based tiles is crucial, coupled with the use of available renewable energy sources. According to this viewpoint, the purpose of this study is to determine the power optimization (PO) using tile deployment in treadmills. This power can be utilized to power up the treadmill and other equipment of the gym.  In accordance with the frequency of movement among the various places of gym and equipment more energy can be generated to power up the gym equipment. In the study Linear regression (LR) machine learning techniques employed for power optimization of treadmill. These techniques are used to optimize tile power in relation to sensor quantity and subject weight. The data was acquired using the gym. Two alternative statistical criteria (MSE and accuracy) are discussed in the paper to assess the efficacy of these algorithms. Model accuracy was expected to range from 0.93 to 0.9946 by the MSE of models. In conclusion, the present paper is reporting that the LR algorithm can be used for the tile deployment on a suitable place of gym. The tile deployment was also anticipated by the LR algorithm, and as a result, the tile could be placed on the treadmill because the frequency of movement is more as compared to other places.