NHS England’s modelling for the Long Term Workforce Plan

NHSE then used the Python model to test the impact of additional policy interventions that might reduce the shortfall by 2036-37, the end of the modelling period. [...] NHSE analysts told us that the complexity of the manual adjustments in the triangulation process meant that this component of the modelling pipeline could not be brought into the Python model within the timetable they had been given. [...] The manual adjustments made in the final components of the modelling pipeline and manual transfers of data introduce a risk of model data and assumptions being inconsistent across the modelling pipeline and increase the likelihood of error. [...] We were not able to attempt to replicate the triangulation process and test whether the calculations in the model were implemented as intended due to the lack of technical documentation and the complex and undocumented manual adjustments made in this part of the modelling pipeline. [...] 15 However, this first version of the modelling pipeline as a whole has significant weaknesses, including the lack of integration between different parts of the pipeline and the manual adjustments to balance supply and demand gaps in the triangulation models.