<p>The increasing electrification of aeronautic transportation and extraterrestrial habitation has resulted in a need for cost-effective and preferably non-destructive insulation health diagnostic solutions. One promising technique that may predictively assess insulation health is machine learning (ML)-assisted quantitative colorimetry. It is known that polymers discolor as they undergo environmentally driven oxidation/crosslinking heretofore referred to as ‘aging’, however quantitative correlation of color to key insulation performance metrics of interest remains in its infancy. Within this work, two commercially relevant polymers were thermally aged at three different temperatures and mechanical, colorimetric, chemical, and dielectric diagnostics were tracked as a function of aging duration. A neural network ML algorithm was trained using coupon-level generated data to predict diagnostic values based on polymer ID and aging condition inputs, with good fidelity observed between predicted and actual values.</p> Graphical abstract <p></p>