Machine learning and XAI approaches highlight the strong connection between O 3 and N O 2 pollutants and Alzheimer's disease

Abstract

Alzheimer's disease (AD) is the most common type of dementia with millions of affected patients worldwide. Currently, there is still no cure and AD is often diagnosed long time after onset because there is no clear diagnosis. Thus, it is essential to study the physiology and pathogenesis of AD, investigating the risk factors that could be strongly connected to the disease onset. Despite AD, like other complex diseases, is the result of the combination of several factors, there is emerging agreement that environmental pollution should play a pivotal role in the causes of disease. In this work, we implemented an Artificial Intelligence model to predict AD mortality, expressed as Standardized Mortality Ratio, at Italian provincial level over 5 years. We employed a set of publicly available variables concerning pollution, health, society and economy to feed a Random Forest algorithm. Using methods based on eXplainable Artificial Intelligence (XAI) we found that air pollution (mainly $O_3$ and $N{O}_2$ ) contribute the most to AD mortality prediction. These results could help to shed light on the etiology of Alzheimer's disease and to confirm the urgent need to further investigate the relationship between the environment and the disease.

Keywords: Alzheimer; Explainable artificial intelligence; Machine learning; One health; Pollution; Standardized mortality ratio (SMR).

Figure 1

Flowchart of the implemented analysis. After a data collection and a pre-processing phase, we selected the best model (among LM and RF) to forecast the SMR for the 107 Italian provinces between 2015 and 2019. Then, we developed a feature importance procedure to improve the performance of the selected algorithm and to measure the importance of each variable in the model.

Figure 2

Panel (A) shows the average standardized mortality rate distribution for Alzheimer’s disease within Italian provinces; Panel (B) and Panel (C) show the average average distribution of $O_3$ and $N{O}_2$ concentrations ($\mu g/m^3$) at Italian provincial level. This image has been created with the software package “sf” of R 4.2.2.

Figure 3

MAE for the three implemented models: linear model (LM), random forest (RF) and random forest with boruta (RF+B). Each distribution was computed through a 5-fold cross validation procedure repeated 100 times.

Figure 4

Average feature importance obtained through RF model after a 5-fold CV procedure with 100 ripetitions, for the considered time span (2015–2019). The white boxes show that the corresponding feature was rejected in the feature selection procedure relied on the Boruta algorithm. Panels (A) and (B) are referred to RF models trained with and without centrality network features respectively.

Figure 5

SHAP distribution values of the most influential features for the considered time window (2015–2019). Each point in the same row corresponds to a different province.

Figure 6

Average map between the SMR distribution and the $O_3$ and $N{O}_2$ concentrations shown in Fig. 2. Before averaging the 3 distributions, minimum maximum normalisation of the values was performed. This image has been created with the software package “sf” of R 4.2.2.