Leveraging machine learning for effective child maltreatment prevention: A case study of home visiting service assessments.

BACKGROUND: Early identification of children and families who may benefit from support is crucial for implementing strategies that can prevent the onset of child maltreatment. Predictive risk modeling (PRM) may offer valuable and efficient enhancements to existing risk assessment techniques. OBJECTIVE: To evaluate the PRM's effectiveness against the existing assessment tool in identifying children and families needing home visiting services. PARTICIPANTS AND SETTING: Children born in hospitals affiliated with the Bridges Maternal Child Health Network in Orange County, California, from 2011 to 2016 (N = 132,216). METHODS: We developed a PRM tool by integrating a machine learning algorithm with a linked dataset of birth records and child protection system (CPS) records. To align with the existing assessment tool (baseline model), we limited the predicting features to the information used by the existing tool. The need for home visiting services was measured by substantiated maltreatment allegation reported during the first three years of the child's life. RESULTS: Of the children born in Bridges Network hospitals between 2011 and 2016, 2.7 % experienced substantiated maltreatment allegations by the age of three. Within the top 30 % of children with high-risk scores, the PRM tool outperformed the baseline model, accurately identifying 75.3 %-84.1 % of all children who would experience maltreatment substantiation, surpassing the baseline model's performance of 46.2 %. CONCLUSIONS: Our study underscores the potential of PRM in enhancing the risk assessment tool used by a prevention program in a child welfare center in California. The findings provide valuable insights to practitioners interested in utilizing data for PRM development, highlighting the potential of machine learning algorithms to generate accurate predictions and inform targeted preventive services.

Children's bullying victimization and maternal suicidal ideation among multicultural families in South Korea: Heterogeneity by family socioeconomic status.

RATIONALE: Children from multicultural families in Korea are vulnerable to bullying victimization. Despite growing evidence on the mental health consequences of bullying victimization for victims, little is known about the spillover effects on their immigrant mothers. OBJECTIVE: This study examined the association between children's bullying victimization and their immigrant mothers' suicidal ideation. The potential moderating role of family socioeconomic status was also investigated. METHODS: Using nine waves of the Multicultural Adolescents Panel Study spanning from 2011 to 2019 (N = 1466), this study estimated individual fixed effects models to control for unobserved individual-level heterogeneity. Interaction models were used to investigate potential heterogeneity by family socioeconomic status, including maternal education, household income, and maternal occupational status. RESULTS: Fixed effects estimates revealed that children's bullying victimization is associated with an increased likelihood of suicidal ideation among marriage migrant mothers (b = 0.012, p < 0.05), even after controlling for unobserved time-invariant confounders as well as a set of time-varying covariates. Family socioeconomic status moderated this association. The association between children's bullying victimization and immigrant mothers' suicidal ideation was stronger for those with low levels of education and household income. No such moderating effects were observed for maternal occupation. CONCLUSIONS: The findings of this study suggest that efforts to address the mental health consequences of bullying victimization among multicultural family children should extend beyond the victim to their immigrant mothers. When developing interventions to reduce suicidal ideation among immigrant mothers whose children have been victimized, policymakers may wish to consider the moderating role of family socioeconomic status.

Predictors of Maternal Recidivism in the Child Protection System.

Child maltreatment recidivism is typically measured and studied at the individual level. Conditions that give rise to child abuse and neglect, however, typically affect multiple children in a given family. In the current study, we estimated maltreatment recidivism at the maternal level and examined its risk as a function of maternal sociodemographic characteristics that may change over time. Using linked administrative records, we identified a subset of first-time mothers in California whose first child was reported to the child protection system (CPS) between birth and age 5 and who then gave birth to another child (n = 14,715). Following the firstborn child's CPS reporting, nearly half of these mothers (43.3%) were re-reported concerning the non-firstborn children during the first 5 years of the child's life. Risk factors consistently documented across births were associated with a heightened risk of maternal CPS recidivism. Our study advances an understanding of the full extent of maltreatment recidivism by broadening the focus from individual children.

A Population-Based Study of Mental Health Diagnoses and Child Protection System Involvement Among Medicaid-Insured Children.

OBJECTIVE: To determine the population prevalence of diagnosed mental health disorders among Medicaid-insured children <18 years old in California based on levels of current and past child protection system (CPS) involvement. STUDY DESIGN: In this retrospective, population-based study, we examined the full population of children enrolled in California's Medicaid program for at least 1 month between 2014 and 2015 and who had at least 1 claim during that period (n = 3 352 886). Records for Medicaid-insured children were probabilistically linked to statewide CPS records of maltreatment and foster care placements since 1998. A primary or secondary mental health diagnosis was classified using International Classification of Diseases codes. RESULTS: Overall, 14% (n = 470 513) of all children insured through Medicaid in 2014-2015 had a documented mental health diagnosis. Among children with a diagnosis, the percentage with CPS involvement (ie, any report for maltreatment) was nearly twice that of the Medicaid population overall (50.4% vs 26.9%). This finding held across all diagnostic groups but with notable variations in magnitude. A graded relationship emerged between the level of CPS involvement and the likelihood of a mental health diagnosis. Diagnoses among children reported for maltreatment were common, regardless of placement in foster care. CONCLUSIONS: Findings document high rates of both mental health diagnoses and past child protection involvement in a population of Medicaid-insured children. Most children reported for maltreatment will never be placed in foster care, underscoring the importance of ensuring that the children who remain at home receive the proper array and coordination of services.

Predicting youth at high risk of aging out of foster care using machine learning methods.

BACKGROUND: Youth who exit the nation's foster care system without permanency are at high risk of experiencing difficulties during the transition to adulthood. OBJECTIVE: To present an illustrative test of whether an algorithmic decision aid could be used to identify youth at risk of existing foster care without permanency. METHODS: For youth placed in foster care between ages 12 and 14, we assessed the risk of exiting care without permanency by age 18 based on their child welfare service involvement history. To develop predictive risk models, 28 years (1991-2018) of child welfare service records from California were used. Performances were evaluated using F1, AUC, and precision and recall scores at k %. Algorithmic racial bias and fairness was also examined. RESULTS: The gradient boosting decision tree and random forest showed the best performance (F1 score = .54-.55, precision score = .62, recall score = .49). Among the top 30 % of youth the model identified as high risk, half of all youth who exited care without permanency were accurately identified four to six years prior to their exit, with a 39 % error rate. Although racial disparities between Black and White youth were observed in imbalanced error rates, calibration and predictive parity were satisfied. CONCLUSIONS: Our study illustrates the manner in which potential applications of predictive analytics, including those designed to achieve universal goals of permanency through more targeted allocations of resources, can be tested. It also assesses the model using metrics of fairness.

Cumulative Rates of Child Protection Involvement and Terminations of Parental Rights in a California Birth Cohort, 1999-2017.

Objectives. To document the cumulative childhood risk of different levels of involvement with the child protection system (CPS), including terminations of parental rights (TPRs).Methods. We linked vital records for California's 1999 birth cohort (n = 519 248) to CPS records from 1999 to 2017. We used sociodemographic information captured at birth to estimate differences in the cumulative percentage of children investigated, substantiated, placed in foster care, and with a TPR.Results. Overall, 26.3% of children were investigated for maltreatment, 10.5% were substantiated, 4.3% were placed in foster care, and 1.1% experienced a TPR. Roughly 1 in 2 Black and Native American children were investigated during childhood. Children receiving public insurance experienced CPS involvement at more than twice the rate of children with private insurance.Conclusions. Findings provide a lower-bound estimate of CPS involvement and extend previous research by documenting demographic differences, including in TPRs.Public Health Implications. Conservatively, CPS investigates more than a quarter of children born in California for abuse or neglect. These data reinforce policy questions about the current scope and reach of our modern CPS.

The Children's Data Network: Harnessing the scientific potential of linked administrative data to inform children's programs and policies.

The Children's Data Network (CDN) is a data and research collaborative focused on the linkage and analysis of administrative records. In partnership with public agencies, philanthropic funders, affiliated researchers, and community stakeholders, we seek to generate knowledge and advance evidence-rich policies that improve the health, safety, and well-being of the children of California. Given our experience negotiating access to and working with existing administrative data (and importantly, data stewards), the CDN has demonstrated its ability to perform cost-effective and rigorous record linkage, answer time-sensitive policy- and program-related questions, and build the public sector's capacity to do the same. Owing to steadfast and generous infrastructure and project support, close collaboration with public partners, and strategic analyses and engagements, the CDN has promoted a person-level and longitudinal understanding of children and families in California and in so doing, informed policy and program development nationwide. We sincerely hope that our experience-and lessons learned-can advance and inform work in other fields and jurisdictions.

Bacterial pathogens and flora isolated from farm-cultured eels (Anguilla japonica) and their environmental waters in Korean eel farms.

The purpose of this study was to investigate bacterial pathogens and flora in both sick and clinically healthy eels, Anguilla japonica, and the environmental rearing waters of Korean eel farms. Between 2003 and 2010, a total of 621 sick eels were submitted for diagnosis, while 216 healthy eels and 87 environmental water samples were collected during a survey of 26 eel farms in Korea. Seven different bacterial species were obtained from 183 isolates, which were recovered from the internal organs of the 621 sick eels. The most frequently isolated bacterium was Edwardsiella tarda (71.0%), followed by Aeromonas hydrophila (9.3%), Citrobacter freundii (7.7%), Aeromonas veronii (6.0%), Listonella anguillarum (2.7%), Plesiomonas shigelloides (2.2%), and Pseudomonas anguilliseptica (1.1%). From the eel and water samples of the survey, a total of 472 isolates from 34 different species belonging to 15 genera of bacteria were isolated. The most prevalent genus of bacteria was Aeromonas spp. (141/472, 29.8%). Among the 34 types of bacterial species, C. freundii (20.1%) and A. hydrophila (19.9%) were the most frequently isolated. The results of this study indicate that a wide range of bacterial species, which can act as primary or opportunistic pathogens, may be recovered from clinically healthy eels and rearing waters. This study provides baseline information about bacterial pathogens and floral contamination for the control and treatment of bacteria in Korean eel farms.

Characterization of Edwardsiella tarda isolated from farm-cultured eels, Anguilla japonica, in the Republic of Korea.

We surveyed the occurrence of edwardsiellosis on eel farms and investigated the characteristics of Edwardsiella tarda isolated from farm-cultured eels in the Republic of Korea. The occurrence rate of edwardsiellosis was 72% in the investigated samples. Among the edwardsiellosis cases, 46% were found to be mixed infections, with parasites and other kinds of bacteria. Some of the biochemical characteristics of the E. tarda isolates were different from those of the previously reported E. tarda isolated from several kinds of fish from different countries, especially in terms of hydrogen sulfide and indole production. The E. tarda isolated from the eels in the Republic of Korea had the characteristics of two biogroups, the wild-type biogroup and biogroup 1. The enzymatic activity of the E. tarda showed similar patterns to previously reported E. tarda strains and ATCC strains. This is the first it has been reported that E. tarda isolated from farm-cultured eels had some different biochemical characteristics from those of previously reported E. tarda isolated from several kinds of fish.