Theoretical models for food drying commonly utilize an effective diffusivity solved through curve fitting based on experimental data. This creates models with limited predictive capabilities. Multiscale modeling is one approach which can help transition to a more physics-based model minimizing the empirical information required while improving a model’s predictive capabilities. However, to enable an accurate scaling operation, multiscale models require diffusivity at a fine scale (microscale). Measuring these properties is experimentally costly and time consuming as they are often temperature and/or moisture dependent. This research conducts an inverse analysis on a multiscale homogenization food drying model to deduce the temporal diffusivity of intracellular water. A representation of the real cellular water breakdown was considered and appropriate assumptions to represent its cellular heterogeneity, in relation to time, were investigated. The work uncovered that a linear decrease in intracellular water content could be assumed and thus a function for its diffusivity was developed. The proposed function is in terms of sample temperature and intracellular water content opening the possibilities to be applied to various food materials.