The sympathetic nervous system (SNS) is essential for the body's immediate response to stress, initiating physiological changes that can be measured through sympathetic nerve activity (SNA). While microneurography (MNG) is the gold standard for direct SNA measurement, its invasive nature limits its practical use in clinical settings. This study investigates the use of multi-wavelength photoplethysmography (PPG) as a non-invasive alternative for SNA measurement. Key features are extracted from the pulsatile components of red and green PPG signals to train a linear regression machine learning (ML) model to predict R-wave-triggered spike count (SPR), a biomarker derived from MNG. The study correlates PPG-derived features with ground truth SPR to develop a predictive model capable of detecting SNA during induced physical stress (isometric handgrip and cold pressor) and cognitive stress (mental arithmetic and Stroop test). Unlike previous research that relies on subjective stress indicators, our work utilizes MNG-derived SPR as an objective ground truth for validation. Our findings demonstrate strong agreement between PPG-predicted SPR values and those obtained via MNG, with red PPG showing a higher correlation. The green wavelength PPG exhibits greater sensitivity in detecting stress-induced SNA, particularly during stress onset, where it outperforms the MNG method in capturing immediate responses to stressors such as mental arithmetic and the cold pressor task. To the best of our knowledge, this is the first study to directly compare PPG-derived SNA estimates with MNG, offering a promising pathway for developing wearable, non-invasive tools for continuous stress monitoring and sympathetic arousal detection.