<p>The key requirements for medical implantable devices are low energy consumption and reliable computational precision. In this context, we design a novel filter for ECG signal denoising using the stochastic computing (SC) method. Compared to traditional binary methods, SC offers lower power consumption and reliable precision. Previous stochastic circuits use independent random number sources (RNS) to generate stochastic numbers (SN). In complex computing scenarios, multiple SNs require multiple groups of independent RNSs, leading to significant resource overhead, which negates the low power advantage of SC. Therefore, we propose a new shareable RNS that combines a nonlinear function with an LFSR. This RNS generates multiple SNs with low auto-correlation, which can be isolated from each other by simple de-correlation methods (such as inserting a delay unit). Additionally, improving precision substantially increases computation time, resulting in higher power consumption. Thus, we apply a low-delay stochastic multiplier to ECG signal processing to reduce computation time and power consumption of the implantable device.</p>