We investigate the combined temperature and magnetic field dependent behavior of high NV-density microdiamonds in an all-optical frequency-domain based setup, utilizing the change in fluorescence lifetime. We use these frequency-domain data to train neural networks for the prediction of temperatures and magnetic fields. At zero magnetic field, we show the ability to predict temperatures in the range of 0°C to 100°C, with a standard deviation of 1.24°C. Furthermore, we show the ability to predict applied magnetic fields, while varying the ambient temperature, with a higher accuracy, compared to the approach of observing of the fluorescence intensity at a single excitation frequency.