<p>Machining processes are largely reliant on manual intervention and non-value-added processes, such as post-process inspection, to achieve end-product conformance. However, the ever-increasing demand for high manufacturing productivity combined with low costs and high product quality requires online monitoring systems to provide real-time insights into the cutting process and minimize the volume of non-value-added processes. Most of the published work on machining process monitoring focuses on intrusive measurement equipment, such as dynamometers, to predict the dimensional quality of machined parts, preventing industrial exploitation due to practical limitations. The main focus of this work is to address this issue by developing a new product health monitoring method for machining processes using non-intrusive and low-cost instrumentation and data acquisition (DAQ) hardware. The sensing setup in this work includes an acoustic emission (AE) sensor and two accelerometers in the work holding. The proposed monitoring system is applied to milling experiments using Gaussian process regression (GPR) for probabilistic nonlinear in-process product condition monitoring. Validation results show the effectiveness of the GPR model to provide accurate probabilistic predictions of product health metric deviations with reasonable uncertainty estimates considering the large variability of the data. In addition, a Bayesian inference methodology is derived to dynamically incorporate subsequent information from on-machine probing (OMP) with a comparator method, improving the accuracy and robustness of the proposed solution. Specifically, it is demonstrated that a precision-weighted combination of prior information from the posterior predictive distribution for a future observation and new metrological information from on-machine comparator measurement (OMCM) can clearly improve posterior inferences about the end product condition.</p>