BACKGROUND
Many studies have investigated the link between exposure to several air pollutants such as carbon dioxide (CO2), carbon monoxide (CO), and ozone (O3) and how they impact the overall completed suicide rate. However, It is unclear how air pollution may affect the increased suicide rate. Furthermore, it is unclear which pollutants may lead to an increased completed suicide rate.
OBJECTIVE
Therefore, our study aimed to predict the rise in suicide rates using publicly available open air pollution data.
METHODS
Public Organization for Economic Co-operation and Development (OECD)data was used to extract relevant air pollutants data. Our study included countries of Turkey (2019), Greece (2019), Slovak Republic (2019), United Kingdom (2019), Canada (2019), Luxembourg (2019), Poland (2019), Japan (2019), Hungary (2019), Lithuania (2020), Italy (2017) Spain (2020) Portugal (2018), Ireland (2018),Denmark (2018), Germany (2020), Chile (2018), Netherlands (2020), Canada (2019), Austria (2020) Czech Republic (2020), Switzerland (2018), Australia (2020), Sweden (2018), Iceland (2020), United States (2020), Finland (2018), Latvia (2020) Belgium (2018), Estonia (2020), Slovenia (2020), and Korea (2019).
Air pollutants of CO2, CO, greenhouse gas (GHG), Nitrogen oxides (NOX), Sulfur dioxide (SOX), and Volatile organic compounds (VOC) were used to assess the air quality of the countries mentioned above. We have divided countries into high and low-suicide-risk countries according to the mean completed suicide rate of the measured countries.
RESULTS
According to our main study results, the mean suicide rate among countries was 11.36 per 100.000 people. A significant correlation was not found between the suicide rates and levels of measured air pollution parameters. The correlation was negative only between SOX and the suicide rate.
Random Forest (RF) Machine Learning (ML) model showed that we predicted a low or high suicide rate status with an accuracy of 90% (area under curve (AUC): 94%). The level of CO2 was the most predictive factor according to the RF model.
Furthermore, the Neural Networks (NN) ML model predicted a low or high suicide rate with an accuracy of 90% (AUC: 75%). The level of CO was the most predictive factor according to the NN model.
CONCLUSIONS
In conclusion, certain levels of air pollutants may be used to predict the completed high or low suicide rate status.
CLINICALTRIAL
-