Background

Aggressive behaviors in psychiatry represent a fundamental health problem that impact patients' and caregivers' lives, as well as the whole social system. Aggression can be considered in psychiatry as a transdiagnostic dimension, not necessary linked to a specific diagnostic entity. It refers primarily to two different types of aggressive behaviors:

• towards others; and
• towards themselves (suicidal behaviors).

Aggressive behaviors towards others in patients with psychiatric disorders are increasing in the last decades.

In parallel, according to the World Health Organization, more than 700,000 people die due to suicide every year. For every suicide, there are likely 20 other people who attempt suicide (www.who.int).

Many overlapping clinical and environmental risk factors have been implicated for aggressive and suicidal behaviors. They have both been shown to be associated with many clinical and psychopathological dimension such as depression, psychosis substance abuse, and impulsivity, as well as environmental risk factors, including family disfunction and childhood trauma. Blood parameters have also been involved in these behavioral disturbances. In the last years, neuroimaging studies have generated a large body of knowledge regarding brain morphological abnormalities in patients with psychiatric disorders. Interestingly, current advances in the field have focused on the need for more precise neuroimaging biomarkers for aggressive and suicidal behaviors. Specifically, converging evidences found brain structural abnormalities in cortical and subcortical volumes in patients presenting with aggressive and suicidal behaviors. In particular, specific alterations have been shown in frontal and prefrontal cortices and in limbic structures, including amygdala and hippocampus. Further information may derive form considering the anatomical complexity of the hippocampus and analyzing the volumetric differences at its subdivision in subfields (See preliminary data). In the everyday clinical practice, during the psychiatric examination, the analysis of physical behaviors and facial expressions of patients is of particular importance for clinical diagnoses and severity assessment. This includes the early detection of alarming behaviors. Recently, quantitative methods have been developed for the extraction of human movements from video records, using advanced artificial intelligence (AI) and physical models. Specifically, convolutional neural networks (CNNs) have achieved excellent results in distinguishing between aggressive and non-aggressive behaviors from video data. These data, together with automated system for identifying human emotions from facial expressions, may contribute to an early identification of aggressive and suicidal behaviors. Previous studies have looked into predicting aggressive/suicidal behavior. Although different approaches exist, today there are no prediction models taking into account all the variables mentioned above. Furthermore, prediction models have not been articulated in actionable tools applied in the real world. Therefore, an accurate early identification of aggressive and suicidal behaviors through ad-hoc software products is absolutely needed and may contribute to saving a higher number of lives. The economic and social benefits will be great given the impact of behavioral alternations in psychiatric patients (See fourth paragraph).

Specific Aim of the project

The aim of this project is to develop a software platform capable of predicting – with an unprecedented level of precision and accuracy – aggressive/suicidal behaviors in psychiatric patients, by integrating multimodal information from clinical variables, environmental factors, blood parameters, volumetric neuroimaging data, and motion/facial expression data from video records. The software should be able to generate a risk score for aggressive/suicidal behaviors in real time, according to the variables provided. The tool will recommend specific actions based on risk level.

The project will consist of three parts:

• A fundamental research study;
• An industrial research study; and
• An experimental development study.

The project aims at reaching for a technology maturity level of TRL 7 for the developed software.

Working group

Dott.ssa Delfina Janiri - Responsabile scientifico UCSC

Prof.ssa Simona Gaudino

Prof. Gabriele Sani

Prof. Marco De Spirito

Prof.ssa Silvia Giovannini

Prof.ssa Daniela Pia Rosaria Chieffo

Prof. Giuseppe Maulucci

Prof. Luca Indovina

Prof.ssa Silvia Giovannini

Dott. Federico Tonioni

Dott. Marco Di Nicola

Dott. Gino Pozzi

Dott. Lucio Rinaldi

Dott. Alain Maria Dell'Osso

Procedure di selezione

Procedure chiuse:

535_25_JANIRI_verbale_RIUNIONE IN PRESENZA.pdf

Prot.n.53525_04062025 Bando procedura selezione consulenze FISA 2022

Allegato A.