There are over 40,000 suicide deaths per year in the United States. Predicting those at highest risk of suicide and offering those people extra and just in time support is a priority for the field. Suicide prediction models work by modeling the risk factors a patient may have (e.g., age, access to weapons, gender, etc.) and attempting to identify those at the highest risk. However, in a 2019 JAMA Psychiatry paper, the authors showed that while these models can cluster risk well, the predictive validity associated with a positive result for suicide mortality was extremely low, less than 1% for most models. Thus, these predictive models are not useful when used in isolation.