In this paper, we demonstrate the ability to run an adaptation on field data for a steady-state multiphasepipe flow model for calculating the bottomhole pressure of an inclined well using the deviation survey data, wellhead parameters (pressure, phases flowrates and properties), and the temperature distribution along thewellbore. In the underlying model published earlier in different modifications (Baryshnikov et al., 2020;Kanin et al., 2019), we divide the well into segments and perform the hydraulic calculation of the pressuredistribution along the wellbore via the marching algorithm in which the set of PVT correlations and thesegment model are utilized. The distinguishing feature of the utilized approach is a data-driven segmentmodel required for calculating the pressure drop along the chosen part of the pipe. This model is basedon Artificial Neural Networks (ANNs), which firstly trained on the synthetic data and lab measurements, and after that, additionally tuned on the real field data collected on two wells during the flowback periodconducted via the free-flow production method. During the optimization of the ANNs on the field data, weimply the quasi-steady-state approximation of the multiphase flow, i.e. the transient flow is considered as asequence of the stationary states. In work, we also carry out the comparison of the prediction capability of theconstructed pipe segment model with several most common multiphase flow correlations and mechanisticmodels (within the frame of the marching algorithm) and demonstrate its more accurate outcome. As aresult, we develop a method for calculating the flowing bottomhole pressure value adaptable on the requiredfield data. The model adapted on field data allows estimating the flowing bottomhole pressure at eachtime moment (assuming the quasi-steady-state flow) which is particularly important in the absence of thebottomhole pressure gage at the well.