Using machine learning to identify most salient factors impacting disordered eating risk and treatment among rural adolescents

The National Institute of Mental Health awarded Central Michigan University $558,960 on August 15, 2026, under the Mental Health Research Grants program (CFDA 93.242) to use machine learning to identify salient factors impacting disordered eating risk and treatment among rural adolescents. The project runs through August 14, 2029, and is performed in Mount Pleasant, Michigan.

The research addresses disordered eating prevalence among rural adolescents, which occurs at approximately double the rate found in nationally representative samples. Rural adolescents face unique social and environmental barriers that both increase disordered eating risk and reduce treatment access. The project will examine risk factors, protective factors, and treatment obstacles specific to the rural context—including food insecurity, reduced treatment availability, and mental health stigma—as well as potentially protective factors such as social connectedness and body functionality appreciation. The study will employ machine learning methods to identify which factors most significantly influence disordered eating risk and clinical outcomes in this population, with implications for prevention and early intervention strategies during the critical developmental period of adolescence.

Abstract

Disordered eating is a critical public health issue in the united states, due to the alarmingly high prevalence, and myriad of negative physical and psychosocial consequences. Adolescence is a critical developmental period for both the prevention and treatment of disordered eating. Many physical and social changes that occur in adolescence increase disordered eating risk, and early treatment intervention is imperative for treatment prognosis. However, social and environmental differences lead to high prevalence and worse clinical outcomes for disordered eating among rural adolescents. Rural adolescents face rates of disordered eating that are approximately double compared to nationally representative samples. Alarmingly high rates of disordered eating among rural adolescents are likely because they face unique social and structural influences that both increase the likelihood of developing disordered eating but also leads to lower likelihood of treatment. Factors that have been shown to be associated with increased risk of disordered eating in other populations are elevated in rural adolescents but also may be more iatrogenic in rural communities. For example, food insecurity is more common in rural populations, but harm may be further compounded by lack of access to healthful foods in rural communities. Additionally, rural adolescents may be more likely to delay or never receive care because of reduced treatment access and social stigma around mental health that is particularly pervasive in rural communities. IT is also likely that factors such as social connectedness and body functionality appreciation may be uniquely protective for disordered eating in the rural context. However, there has never been a study designed to examine disordered eating risk factors, protective factors, or treatment obstacles specifically within rural adolescent populations. The proposed study would fill a critical GAP in identifying the most salient risk factors, protective factors, and treatment obstacles as well as population-identified solutions within the context of rural adolescents. The study findings can be used to develop culturally relevant prevention and treatment interventions to reduce disordered eating rates among rural adolescents. We will use a concurrent triangulation mixed-methods approach including a cross-sectional survey (n=1,000) among rural high school students that will be analyzed using random forest algorithms, a machine learning technique, and semi-structured interviews (n=40) among rural high school students experiencing disordered eating, who both have and have not received treatment, to achieve the following specific aims: aim 1: determine the most salient risk/protective factors that predict disordered eating among rural emerging adults. Aim 2: identify the most salient obstacles to disordered eating treatment that exist in rural communities and identify population- identified solutions to treatment obstacles.