AI-Enhanced RF Biosensing for Microalgal Biomolecule Characterization

Efficient lipid quantification in microalgae is critical for advancing sustainable biofuel production. However, conventional methods remain invasive, time-consuming, and reliant on complex sample preparation. Here, we present a noninvasive, real-time approach for lipid estimation in Scenedesmus sp. by integrating high-frequency radio frequency (RF) biosensing with machine learning (ML). Microalgae cultures were subjected to nitrogen starvation, and dielectric responses were measured using a coaxial open-ended RF probe connected to a vector network analyzer across the 1–13.6 GHz range. Frequency-dependent variations in reflection coefficient (S11) magnitude and phase were found to correlate with intracellular lipid accumulation over a 23-day period. A supervised ML framework was developed using extracted impedance features, and Random Forest models yielded the highest prediction accuracy. Our results demonstrate that RF-ML integration enables accurate, label-free lipid monitoring, offering a scalable and automated solution for smart bioprocess control. This approach lays the groundwork for intelligent digital twins in algal bioreactors, with broad implications for next-generation bioresource engineering.