Aim Suicide has become a major public health concern in the United States and around the world. Analyzing large datasets is a promising approach to public health research, as artificial intelligence (AI) is becoming important in different applications. Examining suicide rates and their relationships with many factors over a large area should be of significance in addressing prevalent public health concerns. The aim of this study was to identify the overall geospatial landscape of the United States suicide rates and related significant risk factors. Subject and methods Two sets of data were collected from the United States government agencies. The smaller set consists of 10-year (2014–2023) suicide rates in the 50 US states, and the data were used for cluster analysis. The larger dataset consists of 36 variables of the 50 US states. The database covers suicide rates and 35 other variables including percent of adults diagnosed with depression, percent of adults with mental health problems, and alcohol and drug use behavior data; median household income, percent unemployment, percent college education attainment, and population socioeconomic data; and state land area, urban area, and elevation relief physical environmental variables. Statistical correlation and regression analyses were carried out to delineate the geospatial patterns of suicide rates and their relationships to the 35 other variables. Results The 50 US states are classified into four distinct clusters based on suicide rates. Seven of the 35 variables are correlated with suicide rates, with a correlation coefficient r ≥ 0.5. A national linear regression model emerged, with four significant variables at the 5% significance level. The results show that suicide rates are positively related to juvenile (age 10–17 years) suicide rates, percentage of adults with mental health problems, and state elevation relief, but negatively associated with the number of registered psychiatrists. Conclusion Although statistical analyses cannot directly reveal causal relationships between suicide rates and predictive significant variables, the results can be used to facilitate effective screening and mitigation measures. The approach used in this study can be scaled to other national and international geographical settings as future large digital public health databases become available. Comprehensive public health database development, statistical analyses, and machine learning AI would help identify the geospatial hotspots of suicides and formulate timely measures for suicide reduction and prevention.

Analyzing the general suicide landscape of the United States

期刊论文
发表日期：

2026年4月15日

查看全文

X. Mara Chen,
Veera Holdai

摘要

Aim

Suicide has become a major public health concern in the United States and around the world. Analyzing large datasets is a promising approach to public health research, as artificial intelligence (AI) is becoming important in different applications. Examining suicide rates and their relationships with many factors over a large area should be of significance in addressing prevalent public health concerns. The aim of this study was to identify the overall geospatial landscape of the United States suicide rates and related significant risk factors.

Subject and methods

Two sets of data were collected from the United States government agencies. The smaller set consists of 10-year (2014–2023) suicide rates in the 50 US states, and the data were used for cluster analysis. The larger dataset consists of 36 variables of the 50 US states. The database covers suicide rates and 35 other variables including percent of adults diagnosed with depression, percent of adults with mental health problems, and alcohol and drug use behavior data; median household income, percent unemployment, percent college education attainment, and population socioeconomic data; and state land area, urban area, and elevation relief physical environmental variables. Statistical correlation and regression analyses were carried out to delineate the geospatial patterns of suicide rates and their relationships to the 35 other variables.

Results

The 50 US states are classified into four distinct clusters based on suicide rates. Seven of the 35 variables are correlated with suicide rates, with a correlation coefficient r ≥ 0.5. A national linear regression model emerged, with four significant variables at the 5% significance level. The results show that suicide rates are positively related to juvenile (age 10–17 years) suicide rates, percentage of adults with mental health problems, and state elevation relief, but negatively associated with the number of registered psychiatrists.

Conclusion

Although statistical analyses cannot directly reveal causal relationships between suicide rates and predictive significant variables, the results can be used to facilitate effective screening and mitigation measures. The approach used in this study can be scaled to other national and international geographical settings as future large digital public health databases become available. Comprehensive public health database development, statistical analyses, and machine learning AI would help identify the geospatial hotspots of suicides and formulate timely measures for suicide reduction and prevention.