Dementia death rates prediction

Abstract

Background: Prevalence of dementia illness, causing certain morbidity and mortality globally, places burden on global public health. This study primary goal was to assess future risks of dying from severe dementia, given specific return period, within selected group of regions or nations.

Methods: Traditional statistical approaches do not have benefits of effectively handling large regional dimensionality, along with nonlinear cross-correlations between various regional observations. In order to produce reliable long-term projections of excessive dementia death rate risks, this study advocates novel bio-system reliability technique, that being particularly suited for multi-regional environmental, biological, and health systems.

Data: Raw clinical data has been used as an input to the suggested population-based, bio-statistical technique using data from medical surveys and several centers.

Results: Novel spatiotemporal health system reliability methodology has been developed and applied to dementia death rates raw clinical data. Suggested methodology shown to be capable of dealing efficiently with spatiotemporal clinical observations of multi-regional nature. Accurate disease risks multi-regional spatiotemporal prediction being done, relevant confidence intervals have been presented as well.

Conclusions: Based on available clinical survey dataset, the proposed approach may be applied in a variety of clinical public health applications. Confidence bands, given for predicted dementia-associated death rate levels with return periods of interest, have been reasonably narrow, indicating practical values of advocated prognostics.

Keywords: AI; Dementia; Mathematical biology; Public health; Risk forecast.

Fig. 1

World’s countries map with dementia cases. All world’s countries have been accounted for in this study, (https://ourworldindata.org/causes-of-death)

Fig. 2

Dementia percentage in Ontario, Canada, by age groups and gender [34]

Fig. 3

Example of how 2 components, X and Y, being merged to create 1 new synthetic vector $\overrightarrow{R}$. Red ellipse highlights case of simultaneous maxima for 2 different biosystem components

Fig. 4

Dementia, Left: 2D surface of annual death cases as % of local population, per country, and per year. Right: Annual recorded death rate in percent as 195D vector $\overrightarrow{R}$. scaled by in percent of corresponding country’s population

Fig. 5

Death rate prediction. a 100-years return level extrapolation of $p_k\left(\lambda \right)$ towards critical level (marked by star), in percent (%) to local population. Extrapolated 95% CI (Confidence Interval) marked by dotted lines. b prediction by Gumbel method. Percentage of local population being on horizontal axis

Fig. 6

Dementia death global statistics, 2^nd order SODP plot