excel with all analysis

NORTHERN COLLEGE HEALTH SERVICES HOSPITAL VISITS FORECASTING 5

Northern College Health Services Hospital Visits Forecasting

My name

Course

Instructor

Date

Northern College Health Services Hospital Visits Forecasting

Health care organizations experience fluctuations in service utilization that can be very difficult to predict. Models for forecasting hospital visits can be useful in informing evidence based planning of inventories and staffing (Lewis et al., 2011). The relevancy and reliability of plans largely depends on the accuracy of the forecasts. Having a good forecast helps the manager improve accuracy of capacity planning, adjust layout decisions based on expected traffic and proactively manage staffing levels. The objective of the assignment is to forecast the number of visits for November using the Northern College Health Services 2008. Some of the main forecasting models used today are average change, average percent change, confidence interval, moving averages and exponential smoothing forecasting.

Average Change

The average change forecasting approach uses the average of the month to month change to do the future projection (Lewis et al., (2009). The advantage of this methodology is that the forecast cannot be distorted by one value. For the purposes of this forecasting task, only data for Northern College Health Services 2008 will be considered. The forecast month (FM) = average hospital visits + midpoint * average change of month to month change (Lewis et al., 2011). The midpoint is determined by (9+1)/2 because the data points are odd in number. The forecast for month of November is calculated as 34.3 + (5*0) = 34.3. The forecast for the month of December is 34 visits. Complete table for the forecasting is provided in Appendix 1.

Average Percent Change

Just as in the average approach forecasting, this forecasting model also uses the average of the changes from month to month but in percent format (Lewis et al., 2011). When forecasting, the average of monthly change is added to the most recent month. The approach is similar to the average change forecasting approach only that it uses the percent changes from period to period. The forecast for the period in consideration is given by the value of the previous season plus the product of the percentage change and the value for the previous month. The forecast month (FM) = Most Recent Month + (Average Percent Change * Most Recent Month) (Lewis et al., 2011). For the month of November, the forecast will be 39+3.75% (39) where 39 is the visits for the month of October. The forecast gives a value of 42. Complete table for the forecasting is provided in Appendix 2.

Confidence Interval Approach

To use confidence interval forecasting as an approach, the standard deviation and/or standard error is used (Lewis et al., 2011). Predicted value for November is the values between lower limit and upper limit of the 95% Confidence Interval which is 20.68801 to 47.91199. This is approximately between 21 and 48. Complete table for the forecasting is provided in Appendix 3.

Moving Averages

The moving averages forecasting approach first examines the variability in the historical data and then provides a way of eliminating the variability by smoothing the trend (Lewis et al., 2011). For the projection for the hospital visits for the month of November will be found using increasing number of months for the moving average. The forecast for visits in November will be the average for the months of January to October of 2008. That is (39+41+34+39+38+19+28+29+37+39)/10 which is approximately 34 hospital visits. This approach is most applicable where there is a constant observed values with trend or seasonality (Lewis et al., 2011). This approach helps separating out random variations in the observed data. Complete table for the forecasting is provided in Appendix 4.

Exponential Smoothing Forecasting

The exponential smoothing forecasting approach uses a damping factor to smooth the forecast (Lewis et al., 2011). What it does is to accommodate all that the model learned up to the previous forecast as per the most recent demand history (Lewis et al., 2011). The damping factor applied for this case is 0.4. using this factor, the model predicts the value for a month by using the formula =(1-f)*most recent + f*historical observed value for the current period. For our case, this gives (1-0.4)*28.3758+0.4*39 which gives a forecast value of 33.55. Using the exponential smoothing forecasting approach, the forecast visits for November is 34. Complete table for the forecasting is provided in Appendix 5.

Comparison of the forecasts for November

Model	Forecast value for November
Average Change	34
Average Percent Change	42
Confidence Interval Approach	Between 20 and 48
Moving Averages	34
Exponential Smoothing Forecasting	34

Examples of real case application of forecasting in health care

Hospitals and health services organizations use forecasting because high levels of varying characteristics and randomness are inevitable. For example, West China Hospital (WCH) uses time-series forecasting to predict daily outpatient visits (Luo et al., 2017). The University of Virginia Biocomplexity Institute also developed a multiple forecasting model for projecting future hospitalization rates and utilization rates for hospital beds (University of Virginia School of Engineering, 2020). Additionally, Beth Israel Deaconess Medical Center developed a model for forecasting COVID-19 (Stevens et al., 2020).

Researchers also predicted COVID confirmed cases and deaths using a simple exponential smoothing time series model of forecasting (Petropoulos et al., 2020). The researchers argued that their model competitive accuracy in forecast and superior estimates of uncertainty which can be helpful and clinically relevant. Although there are many forecasting models, some organization just forecast trends bsd on judgment and knowledge. For example, Good Judgment Inc uses judgmental forecasting methodology to provide projections that can be used to inform decision making for their clients. Good judgment created a COVID dashboard that forecasts notable socioeconomic milestones with respect to COVID-19. She predicts when and how key developments regarding COVID will be made. For example, they predicted the timing for the development of a vaccine and when normal flighs may resume. However, the accuracy of these judgment-based forecasting depends on the matter being forecasted and the knowledge of the subject.

Conclusion

In conclusion, for the purposes of predicting future hospital visits, the best models are moving averages and exponential smoothing approaches. These two approaches eliminate random variability and hence provide the health services managers with dependable forecast data for extrapolate trends. However, both models are relatively naive and their effectiveness depends on the knowledge and experience of the health services manager to apply the right model. These models can only be used when prudent judgment indicates that historical data can be used to forecast the future.

References

Lewis, J. B., McGrath, R. J., & Seidel, L. F. (2011). Essentials of applied quantitative methods for health services managers. Jones & Bartlett Publishers.

Luo, L., Luo, L., Zhang, X., & He, X. (2017). Hospital daily outpatient visits forecasting using a combinatorial model based on ARIMA and SES models. BMC health services research, 17(1), 1-13. https://bmchealthservres.biomedcentral.com/articles/10.1186/s12913-017-2407-9?optIn=false

Petropoulos, F., Makridakis, S., & Stylianou, N. (2020). COVID-19: Forecasting confirmed cases and deaths with a simple time series model. International Journal of Forecasting. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7717777/

University of Virginia School of Engineering. (2020, December 15). University of Virginia Biocomplexity institute launches national COVID-19 medical resource demand dashboard. Retrieved from https://globenewswire.com/news-release/2020/12/15/2145720/0/en/University-of-Virginia-Biocomplexity-Institute-Launches-National

Stevens, J. P., Horng, S., O’Donoghue, A., Moravick, S., & Weiss, A. (2020, June 29). How one Boston hospital built a COVID-19 forecasting system. Retrieved from https://hbr.org/2020/06/how-one-boston-hospital-built-a-covid-19-forecasting-system

Appendices

Appendix 1: Average Change Forecasting

Month	Hospital visits	Change
Jan	39
Feb	41	2
Mar	34	-7
Apr	39	5
May	38	-1
Jun	19	-19
Jul	28	9
Aug	29	1
Sep	37	8
Oct	39	2
Average	34.3	0
Midpoint		5
Nov forecast		34.3

Appendix 2: Average Percent Change

Month	Visits	Change	% Change
Jan	39
Feb	41	2	5.13%
Mar	34	-7	-17.07%
Apr	39	5	14.71%
May	38	-1	-2.56%
Jun	19	-19	-50.00%
Jul	28	9	47.37%
Aug	29	1	3.57%
Sep	37	8	27.59%
Oct	39	2	5.41%
average			3.79%
FM			40.48

Appendix 3: Confidence Interval Approach

Month	Visits
Jan	39
Feb	41
Mar	34
Apr	39
May	38
Jun	19
Jul	28
Aug	29
Sep	37
Oct	39
average	34.3
std dev	6.945022
Z-Crit	1.960
LCL	20.68801
UCL	47.91199
FM Nov	Between 20.68801 and 47.91199

Appendix 4: Moving Average Forecasting

Month	Visits	n=2	n=3	n=4	n=5	n=6	n=7	n=8	n=9	n=10
Jan	39
Feb	41
Mar	34	40
Apr	39	38	38
May	38	37	38	38
Jun	19	39	37	38	38
Jul	28	29	32	33	34	35
Aug	29	24	28	31	32	33	34
Sep	37	29	25	29	31	31	33	33
Oct	39	33	31	28	30	32	32	33	34
Nov		38	35	33	30	32	33	33	34	34

Appendix 5: Exponential Smoothing Forecasting

Month	Visits	Forecast
Jan	39
Feb	41	#N/A
Mar	34	39
Apr	39	40
May	38	36
Jun	19	38
Jul	28	38
Aug	29	25
Sep	37	27
Oct	39	28
Nov		34