Control Charts For Seasonal Disease Monitoring And Performance Analysis

Seasonal auto-correlated count processes are common in infectious diseasemonitoring process. Monitoring the process and detective unusual behavior aresignificantly important for process improving. IN this paper we used INAR(p) modelto describe seasonal auto-correlated count process, based on which6control chartswere established(c-chart, residual chart, moving average chart and Shewhart chart,CUSUM chart, EWMA chart based on one-step-ahead forecasting errors). Wediscussed the performance of these control charts via simulation studies with differentvalues of parameter and process mean in INAR(1) models and offered advices oncontrol chart selection in different situation. The results showed that the c-chartperforms better if process shifts and levels of are large, while the moving averagechart is more suitable for cases where process shifts and values are both small. It isnoted that in some situations, the residual chart performs best among first three chartswe mentioned above. As for the three control charts based on one-step-aheadforecasting errors, CUSUM and EWMA perform better than Shewhart chart when and process shifts are relatively small while Shewhart chart is better when the processshifts and are large. What is worth to be noticed is that CUSUM performs morestable than EWMA chart in most cases. Then an example about cases of tuberculosiswas given, in which we used the six control charts to monitor number of patientsbased on observed data from a hospital and found that the performance aresatisfactory.