Abstract/Description
Background: In this study, a new analytical approach was introduced to answer specific questions related to mortalities due to cerebrovascular diseases, heart diseases, and the association of these mortalities with twelve other causes of death (COD).
Methods: A multivariate time series forecasting model was developed utilizing each of the CODs by taking the weekly and yearly seasonality into account, and the mortality counts were forecasted using the most recent CDC weekly mortality count data. A new COD data matrix was structured for all CODs as a function of weeks by combining the observed and predicted values of the mortality counts. Two new concepts, Weekly Exceedance in Mortality Count (WEMC) Weekly Change in Mortality Indicator (WCMI) were introduced, which are instrumental in computing probabilities of various simple, compound, and conditional events relating to the CODs. To check the validity of the approach, Wilcoxon rank sum test and correlation accuracy were used.
Results: The correlation accuracy was found to be approximately 88% ensuring a good alignment of forecasted mortality values with the observed values. The Wilcoxon rank sum test produced a non-significant p-value of 0.25 confirming the consistency of the analytical procedure. Probabilities associated with WCMIs were also computed as an illustration of the analytical method.
Conclusion: By utilizing this analytical approach, researchers and policymakers not only can gain valuable insights into trends and patterns in mortality but also can compute the likelihood of any number of desired events related to CODs which can be helpful for public health interventions and resource allocation.
College/School/Affiliation
Joint School of Public health
Included in
Community Health and Preventive Medicine Commons, Data Science Commons, Epidemiology Commons, Health Services Research Commons, Statistics and Probability Commons
A Modern Analytical Method to Forecast Cerebrovascular Diseases (CD), and Heart Diseases (HD) Using Multivariate Time Series Model Utilizing the CDC Provisional Mortality Data
Background: In this study, a new analytical approach was introduced to answer specific questions related to mortalities due to cerebrovascular diseases, heart diseases, and the association of these mortalities with twelve other causes of death (COD).
Methods: A multivariate time series forecasting model was developed utilizing each of the CODs by taking the weekly and yearly seasonality into account, and the mortality counts were forecasted using the most recent CDC weekly mortality count data. A new COD data matrix was structured for all CODs as a function of weeks by combining the observed and predicted values of the mortality counts. Two new concepts, Weekly Exceedance in Mortality Count (WEMC) Weekly Change in Mortality Indicator (WCMI) were introduced, which are instrumental in computing probabilities of various simple, compound, and conditional events relating to the CODs. To check the validity of the approach, Wilcoxon rank sum test and correlation accuracy were used.
Results: The correlation accuracy was found to be approximately 88% ensuring a good alignment of forecasted mortality values with the observed values. The Wilcoxon rank sum test produced a non-significant p-value of 0.25 confirming the consistency of the analytical procedure. Probabilities associated with WCMIs were also computed as an illustration of the analytical method.
Conclusion: By utilizing this analytical approach, researchers and policymakers not only can gain valuable insights into trends and patterns in mortality but also can compute the likelihood of any number of desired events related to CODs which can be helpful for public health interventions and resource allocation.
Comments
Keywords:
CDC Provisional Death Counts, Weekly Exceedance in Mortality Count (WEMC), Weekly Change in Mortality Indicator (WCMI), Multivariate Time-series Forecasting Modeling, Cause of Death (COD) Data Matrix