Suicide risk assessment is a process of estimating the likelihood for a person to attempt or die by suicide. The goal of a thorough risk assessment is to learn about the circumstances of an individual person with regard to suicide, including warning signs, risk factors, and protective factors. Risk for suicide is re-evaluated throughout the course of care to assess the patient's response to personal situational changes and clinical interventions. Accurate and defensible risk assessment requires a clinician to integrate a clinical judgment with the latest evidence-based practice, although accurate prediction of low base rate events, such as suicide, is inherently difficult and prone to false positives. The assessment process is ethically complex: the concept of "imminent suicide" (implying the foreseeability of an inherently unpredictable act) is a legal construct in a clinical guise, which can be used to justify the rationing of emergency psychiatric resources or intrusion into patients' civil liberties. Some experts recommend abandoning suicide risk assessment as it is so inaccurate. In addition suicide risk assessment is often conflated with assessment of self-harm which has little overlap with completed suicide. Instead, it is suggested that the emotional state which has caused the suicidal thoughts, feelings or behaviour should be the focus of assessment with a view to helping the patient rather than reducing the anxiety of clinician who overestimates the risk of suicide and are fearful of litigation. In 2017, an example of how to do this in practice was published in the Scientific American. Given the difficulty of suicide prediction, researchers have attempted to improve the state of the art in both suicide and suicidal behavior prediction using natural language processing and machine learning applied to electronic health records.