Development and validation of machine learning models to predict prediabetes using dietary intake data in young adults in Korea: a cross-sectional study.

Pivotal studies, including the Centers for Disease Control and Prevention (CDC)–Kaiser Permanente study focusing on adverse childhood experiences (ACEs), documented cumulative effects of exposure to potentially traumatic experiences (for example, abuse, neglect or violence witnessing) on a wide range of mental and physical health conditions. These discoveries prompted examination of the enduring role of ACEs in health and disease and it is now estimated that the economic burden of exposure to ACEs in the US adult population is US$14.1 trillion annually. Given the accumulating evidence regarding the human and fiscal toll of ACEs, calls to address the prevalence and consequences of ACEs are rapidly increasing. At the forefront of these efforts, California became the first state (in 2020) to implement a publicly supported screening program. Although the goal of the program is to dramatically reduce the burden of ACEs on the population, the value of this public health initiative has been questioned on several grounds, the primary being that ACEs screens identify risk at the population level, yet perform little better than chance at the level of the individual, limiting their ability to direct prevention or intervention resources to the most vulnerable children. There are several potential explanations for why ACEs scores are limited in their ability to better detect health risks. One possibility is that substantial sources of stress and trauma occurring in childhood are missed with existing instruments. Indeed, one early-life exposure that is not currently included in standard screening instruments is unpredictability of parental and environmental signals received by the child, which activates the brain’s stress responses. The concept that unpredictable signals to the developing brain are stressful and disrupt brain maturation arose initially in experimental animal studies. Since then, unpredictable parental care and lack of structure in the family and home environment have been shown to strongly predict poorer cognitive and emotional development across a broad range of cultures and sociodemographic groups. Unpredictability in childhood, independent from parental support and sensitivity, has been linked by several independent groups to decreased executive control, a slower trajectory of cognitive development, poorer memory and increased risk for depression, anhedonia, anxiety and post-traumatic stress disorder later in life. These associations persist after consideration of other well-established ACEs (for example poverty, abuse or neglect), suggesting that unpredictable experiences in themselves are a robust risk factor for poor mental health outcomes, and their absence from existing assessments of ACEs may account for some of the limitations of these assessments in predicting risk profiles. Here, we test the contribution of unpredictability to mental health outcomes in a large, population-based cohort (~30,000 participants). We examine the relative and cumulative contributions of ACEs and unpredictability as risk factors for mental health problems. We leverage the existing ACEs screening implemented in the Children’s Hospital of Orange County (CHOC) primary care network, engaging with families from a broad range of sociodemographic backgrounds. Adding our well-validated five-item measure of unpredictability we address the following questions: (1) When used in routine pediatric primary care, does ACEs screening with the Pediatrics ACEs and Related Life-events Screen (PEARLS) identify children at increased risk of mental health problems? (2) Does screening for parental and environmental unpredictability provide added value in assessing risk beyond ACEs screening?