Skip to main content

Let's Explore: An in-depth review of Predict-Align-Prevent's predictive analytics for child maltreatment prevention.

Predict Align Prevent (PAP) is a non-profit organization dedicated to stopping child maltreatment before it happens. While there is no single solution to end child maltreatment, our approach shifts the conversation upstream to develop a combination of policies, strategies, and resources that aim to prevent child abuse before it happens. The current system is reactionary, only providing outreach and resources after a child enters the system and far too often, missing the cases that end in a child fatality.
In our data-driven three phase approach (predict -> align -> prevent), we first utilize place-based geospatial machine learning to identify where children are at the greatest risk of maltreatment. Data, including child welfare, health, crime, code violations, infrastructure, and more, are analyzed to create a relationship model of maltreatment across space. We validate the model against prior known maltreatment incidents, creating a powerful tool: a predictive maltreatment risk map at scales starting from a few city blocks.
From there, risk and protective factor data is selected, based off research derived from ACEs studies, social determinants of health, and resilience science. We develop a geographic risk and protective factor analysis to determine which risk factors are most harmful and which protective factors are most helpful across each community. The resulting maps show where and which efforts will have the greatest prevention impact.

Through our programs with state child welfare agencies, we have discovered that the highest risk locations for child maltreatment collectively comprise ~10% of the total population in less than ~10% of a city's geography.

This fraction of the population experiences:

  • 60% of child maltreatment
  • 60% of all child fatalities
  • 40% of diagnosed behavioral, conduct, and psychiatric diagnoses in children
  • 70% of violent crimes against elders

A few examples of indicators for child maltreatment that we have identified during our research included: elevated lead levels in children, urinary tract infections, Substance Use Disorder, economic stability, economic supports, acute opioid use, animal abuse, diabetes, self-harm/suicide, asthma, maternal cannabis use, Children's Medicaid, SNAP/ food stamps, assault/community violence, Neonatal Abstinence Syndrome, building and health violations, and the list goes on.

Our model enables targeted prevention and alignment of services to equitably reach the greatest number of those most in need across each community. We have also discovered that ICD 10 health diagnosis codes with known correlation to toxic stress cluster in places with high child maltreatment risk.

These findings validate the theory behind the model: areas with high risk for child maltreatment represent the accumulation of ACEs in the underlying population. Preventing ACEs will prevent maltreatment and a host of other dire, ACEs-related outcomes.

Contact us with questions, comments, or requests for a full presentation of our model:" rel="noopener noreferrer" target="_blank" title="Email Request"> or visit our website: for more information.

Stay tuned for our next article in which we will explore our Align phase!

Add Comment

Comments (0)

Copyright ÂĐ 2023, PACEsConnection. All rights reserved.
Link copied to your clipboard.