Best forex trading account uk
Therefore, when the baseline model did not include utilization measures, such as the demographics model, adding morbidity trajectories would help increase the model performance; on the contrary, when the baseline model included utilization measures, such as the prior cost model, the contribution from adding morbidity trajectories might become minimal due to the overlap of both variables.
In addition, given that many healthcare expenditures are truly random and cannot be foreseen, an upper limit possibly exists on the risk adjustment model's ability to identify high-cost users; the most comprehensive cross-sectional model might have reached the limit so that adding additional information may not have any help.
Finally, the longitudinal information might have been lost during the process of being collapsed into six levels so that adding trajectories did not help the performance of the model. Models with prior costs had much better performance than diagnosis-based models in identifying high-cost users.