Berkelmans G, van der Mei R, Bhulai S, et al. - Based on findings from a large and unbiased sample, researchers herein confirmed previously discovered risk factors for suicide (middle-age, male gender, and unemployment (as measured through benefits)) that remain elevated even when corrected for a wide array of socio-demographic characteristics. Commonly identified protective factors for suicide, like having a higher income or immigration background, were also confirmed to be robust.

• A logistic regression model was fitted on a training set of 5,854 suicides and 596,416 control cases, and then, the performance was evaluated on a test set of 1,425 suicides and 594,893 control cases.

• On accounting the effect of possible correlating risk factors, risk for suicide appeared to be higher for men, middle-aged people, people with low income, those living alone, the unemployed, and those with mental or physical health problems.

• People who were highly educated, had a non-western immigration background, or were living with a partner, were noted to be at lower risk.

• With an AUC of 0.77, the model had a reasonable fit; this is high for a model predicting suicide death and similar to the performance of deep learning models but with the benefit of remaining explainable.