Artificial intelligence may help predict risk factors for suicide in adolescents. That’s what comes out of you Study published in Science Progress, which identified two new predictive factors for suicide risk in children under 12 and highlighted that the use of artificial intelligence can help doctors assess suicide risk. This is a first multidisciplinary study that paves the way for possible new uses of artificial intelligence to support a delicate discipline such as neuropsychiatry.
The study was conducted by doctors and engineers and included the Meyer’s team for child and adolescent psychiatryled by Tiziana Pisano in the Neuroscience Center of Excellence under the direction of Renzo Guerrini, together with colleagues from the joint T3Ddy laboratory – for which Monica Carfagni from the University of Florence and the engineer Kathleen McGreevy from Meyer are responsible – and Giovanni Castellini, associate professor of Psychiatry at the University of Florence.
The retrospective observational study – reports a note – analyzed data from 237 patients hospitalized at Meyer between 2016 and 2020 for suicidal behavior and suicidal ideation.
Aim of the research
The aim is to subsequently identify the first “indicators”, the predictive factors, for the suicide risk of these patients. For each of them, epidemiological and psychopathological data were collected and they were divided into two groups: those who had shown actual suicidal intentions (with a high potential risk to physical health) and those who had instead shown suicidal thoughts. less structured. This is where artificial intelligence and statistics entered the field: The data was organized and analyzed using mathematical-statistical models (“Neural Network Method”, “Random Forest” and “Pearson Chi-Square Test”).
The result highlighted Two new factors were statistically correlated with an increased risk of suicidal behavior in children under 12 years old: one Previous diagnosis of oppositional defiant disorder It is a previous diagnosis of intermittent explosive disorder. In addition, the study highlighted that so-called “suicidal behavior” (i.e. that the patient does not show real suicidal thoughts, but rather triggers a request for help through this behavior) is an important risk factor and has so far been underestimated.
The prediction model developed by the Meyer-Unifi tandem will be able to do this So be an additional tool for detecting early “warning signs” in young and very young people, even in cases previously considered low-risk. And since, as the WHO also reminds us, suicidal behavior is progressive (i.e. suicide is more likely to be completed if previous behaviors or self-harm attempts have already occurred), this tool could Be useful in assisting in the initiation of preventive measures and early therapeutic interventions.
“This first study is very promising because it encourages us to do this “Artificial intelligence may prove to be an additional tool to complement the clinical assessment of patients, which of course can never be replaced.”, explains Tiziana Pisano. “It is essential to have new tools for early assessment of risks to the neuropsychiatric health of adolescents and children, and we know this.” Suicide among the very young is a public emergency: “The data we analyzed shows that the rate of hospitalizations for suicidal behavior and suicidal ideation increased from 27.69 to 45.28% between 2016 and 2020 and, unfortunately, the trend is still increasing.”