October 9, 2023 – h. 10.00

Sala Europa, IRPPS via Palestro 32 – Rome

Abstract

Artificial Intelligence (AI) is transforming people’s life in unprecedented ways. AI models have human or superhuman abilities in multiple tasks, e.g., gaming, driving, conversation, and content organization. In biomedical research, however, AI demonstrated as much promise, e.g., in molecular drug design, as much disappointment, e.g., in clinical drug repurposing or public health intervention. One of the reasons is that the datasets AI feeds on –sourced from real world databases such electronic health records (EHR)– are often littered with bias. Such bias might be irrelevant to predict the happening of health conditions, but it influences any strategy to prevent such conditions from happening. In this talk, we will take a dive into the promises and perils of AI in healthcare, and its troubled relationship with data, bias, and causality. We will explore novel causal AI methodologies able to both provide accurate individual health predictions as well as interventions. Finally, we will present use cases of causal AI on large, integrated EHR data, and an eagle’s view of EHR consortia in the USA.

Short biographies

Mattia Prosperi, PhD, FAMIA, is Professor in the Department of Epidemiology, and Associate Dean of AI and Innovation in the College of Public Health and Health Profession at University of Florida. His background is in computer science engineering, with expertise in machine learning, bio-health informatics, and epidemiology. His research leverages technology and data intelligence to develop prediction and intervention models for improving future health and lives. In his administrative role, his mission is to expand AI infrastructure, training, research and expertise capacity in public health and health professions.

Yi Guo, PhD, is Associate Professor in the Department of Health Outcomes, Policy and Biomedical Informatics, College of Medicine, University of Florida. He has a multi-disciplinary background in the analysis of real-world data, including electronic health records and administrative claims, experimental and observational study design, predictive modeling (e.g., statistical and machine learning), causal modeling, and analysis of patient-reported outcomes in clinical and public health applications, and among various populations, especially vulnerable populations.

Note CNR – IRPPS

Interview with Antonio Tintori

Interview with Giulia Ciancimino

Participants

CNR Press Office: Emanuele Guerrini

Head: Marco Ferrazzoli

Secretariat: Press Office

The Institute for Research on Population and Social Policies of the National Research Council (Cnr-Irpps) has conducted two studies in primary schools in Rome and in upper secondary schools in Italy which show the presence of a pronounced adherence to stereotypical roles in boys and girls and the mitigation of such conditioning in the transition to adolescence.

The data indicate how gaps and stereotypes are still strongly reproduced within the family environment and the important role of the school in countering them.

Three charts are presented below to visualize the results.

• Figure 1. Level of adherence to male and female stereotypical roles
  (responses for children and adolescents):

• Figure 2. Men are more inclined towards:

• Figure 3. Women are more inclined towards: