Motion monitoring technology has been proven to provide great benefits to various areas in digital healthcare such as socially assistive robotics and physical rehabilitation. With the advancement in nanomaterial, low-cost and scalable fabrication of small footprint, high performance stretchable strain sensors became possible, enabling the development of affordable, comfortable, and unobtrusive wearable motion monitoring devices. In the first study, low-cost, ultra-thin piezoresistive sensors have been designed and fabricated utilizing a scalable manufacturing process. A convolutional neural network-based mapping model was developed to establish complex relationships between integrated sensor signals and upper limbs postures, namely joint angles of elbows and shoulders. In addition, sensory input was processed to include temporal information, aimed to address the non-linear and time-varying behavior of the conductive polymer composite-based sensor. An average normalized root mean squared error of 7.13% was achieve, validating the benefit of neural networks in addressing less than ideal sensor behaviors, and the usefulness of low-cost, nonlinear piezoresistive sensors in wearable pose estimation devices. Acknowledging the limitations of the first study, the second study incorporated novel sensor grid to enhance spatial information captured in the sensor array. In addition, a novel data processing pipeline was developed to enhance spatiotemporal understanding, by creating a strain distribution image and employing a novel architecture combining both convolutional neural network and long-short term memory modules. An improved average NRMSE of 4.53% represented a 36% improvement to the first prototype, validating the benefits of the novel sensor grid pattern design.