Sprouting significantly compromises the storage quality of tubers, reducing both shelf life and market value. The ban on traditional chemical inhibitors has led to the adoption of more expensive alternatives, making precise and early intervention increasingly important. In this work, we leverage electrophysiological signals recorded from potato tubers to predict sprouting before any visible signs appear. Our approach combines wavelet-based feature extraction with ML-based regression to estimate the number of days until sprouting. By incorporating uncertainty quantification to filter out unreliable predictions, the system achieves robust performance, enabling accurate forecasts up to 40 days in advance.