Understanding the Point-Spread Function (PSF) morphology is crucial in numerous scientific applications, including astrometry, photometry, and object subtraction. Moreover, accurate knowledge of the PSF facilitates deconvolution, which, in turn, benefits applications such as the study of galaxy kinematics, where disentangling PSF from the galaxy morphology is crucial for accurate extraction of kinematic signatures. Another example is the study of stellar regions like globular clusters or the galactic center, where deconvolution improves astrometry and velocity measurements. It also aids in studying bodies of the Solar System. However, using a PSF extracted directly from observed data can lead to errors due to PSF variations across the field, necessitating a model that generalizes PSF over the field and wavelengths. Modeling a PSF is notoriously difficult, particularly for Adaptive Optics (AO)-assisted instruments like the MUSE Narrow-Field Mode (MFM). We propose a method to fit and predict the scientific PSF for MUSE MFM accurately. Our predictive framework can infer PSF morphology based on weather conditions or engineering data recorded alongside scientific observations. This approach is valuable for retrieving PSFs when absent (such as extragalactic observations). Additionally, this approach can be integrated with forecasted weather data for more precise exposure time estimation.