Using AI to Improve Foster Care Matching & Adoption Outcomes
Updated: Jan 13
A version of this article originally appeared on cdomagazine.tech.
By Thiag Loganathan & Kevin Jones
Nick Garza is all smiles at the courtroom on the day he was officially adopted.
Pic courtesy: Good Morning America
A few months ago, 8-year-old Nike Garza became one happy boy. After spending nearly half of his life (1,553 days to be precise) in the foster care system, he was finally able to join his adoptive family. Extensively captured by the media, the joyous moment was a great display of the State of Texas coming through for its vulnerable citizens.
But one cannot ignore the long wait for Nick to find permanent placement in a loving family he could truly call his own.
How can state child welfare agencies avoid such indefinite delays in finding permanent homes for their clients? A good first step seen in some states has been getting off decades-old legacy systems, bringing down data barriers between agencies and making a strong commitment toward modernization of IT efforts.
We are in a mission to do exactly that, using modern cloud-based AI technologies for the State of Indiana to take steps to adopt a mobile-first, cloud-based approach to leapfrog and go cutting edge, leveraging AI for improved outcomes for children at Indiana Department of Child Services.
AI’s Application in Human Services
Artificial intelligence (AI) has helped make many consumer experiences seamless, intuitive and effortless; it is time for child welfare systems to benefit from its wide-ranging capabilities: automation of mundane tasks, enhanced decision-making abilities, freeing up workers’ time to engage in more impactful work.
Apart from productivity improvements, leveraging chatbots and ease of use examples like NLP/Conversational UI to auto-fill forms, we are looking to operationalize high-impact use cases like "Child at Risk” alerts and "Foster Care Matching".
Flagging Children at Risk
When should a child be separated from their family? This is one of the most difficult decisions for a caseworker to make. Family is the best setting for a child to develop and grow, but you cannot ignore the risk of further abuse, neglect or, worst case, death. The state’s child welfare system kicks into motion when a referral (say, a neighbor or a relative) indicates a child may be maltreated and is suspected to be abused or neglected.
AI can help flag early in the process as when the screening or intake worker takes down details and needs to determine the level of risk a child is in. AI needs training with existing data for training an AI system. The diagram below illustrates some of the data. AI can be more effective in the case of repeat offenders and/or an going support to the family to improve the living conditions for the child, to suggest the next course of action, be it removing the child from the home in light of extended neglect and abuse, or looking at interventions that can tide over child neglect arising from a family’s temporary hardship.
Foster Care Matching and Adoption
In the unfortunate circumstance of separating a child from the family, at times during late hours, AI can be used in finding the right child-foster care provider fit. Using AI, agencies can select the ideal foster care arrangement based on the attributes of the child and the provider.
Further, AI can augment human effort in the recruitment process for adoption by cutting down on time taken to place a child in a loving, permanent home. Advanced ML/AI matching models can help caseworkers with suggestions for the best adoptive family for the child based on their history and application statuses in the past.
This seamless, AI-supported case workers and process should improve overall outcomes for vulnerable children in the state of Indiana.
Operationalizing AI, Security & Adoption
The biggest challenge in operationalizing AI is adoption. It is important to weave the intelligence as recommendations and nudges embedded into the workflows and in context to help case workers take preemptive action. Alerts and notifications can keep caseworkers informed even when they are on the go, while communication between teams can be synced with third-party apps like Gmail, Outlook, Slack and WhatsApp.
Moreover, powerful conversational UI like chatbots and natural language processing (NLP) models for search, dictation and creating narratives on the field ensure a seamless workflow that complements, rather than hinders, a caseworker’s day. This is especially important as many states struggle with alarming attrition rates, and at Indiana there is already precedence in leveraging VR technology to train and onboard the workforce. We look to build on top of that by using AI to democratize knowledge for new caseworkers bringing them up to speed with decades of veteran insights.
However, seamless workflows and actionable intelligence can’t take away from the fact that privacy and security of all data is critical. We are looking to enable Indiana DCS workers with a secure, smart and easy way to use technology to improve outcomes for the children of Indiana, and set an example for others to follow.