AI Risk Scoring for Child Welfare Prevention Cases

What it is for

Risk Assessment & Triage

Scores and ranks children enrolled in prevention services by their predicted likelihood of being the subject of an indicated New York State Central Register investigation within 24 months. Scores are produced approximately 100 days after the start of a new prevention service to inform the FTC Reassessment Planning Conference.

NYC Algorithmic Tools Compliance Report — Family Team Conference Model (FTC)

Run by: Administration for Children's Services (ACS)
Where: No fixed location
Kept: Retained Not specified in the register
Shared with: Accountable organization
Your copy: You cannot see the data it holds about you. What you can do

What it is for

Risk Assessment & Triage

This AI model generates a monthly risk score for children enrolled in prevention services, predicting the likelihood of a child welfare investigation within the next 24 months. The scores are used by senior staff to prioritize which families receive enhanced attention during structured planning conferences held approximately 100 days into a prevention service. The model is supervised — a human specialist reviews and acts on the ranked list; no automated decisions are made. It is built entirely on ACS administrative data and does not collect or disclose any identifying information during prediction.

What it collects and what happens to it

Data taken in

Operational data

Anonymized data

ACS administrative data about prior and current child welfare involvement at the start of a case. This includes New York State Central Register (SCR) investigations and time spent in foster care. Only ACS administrative data are used; no identifying information is collected or disclosed during prediction.

NYC Algorithmic Tools Compliance Report — Family Team Conference Model (FTC)

Processing

Classification & Prediction

The model predicts the probability that a child will be the subject of an indicated NYS Central Register investigation within 24 months from the observation date, producing a ranked list of open prevention cases. Trained on 138,712 historical ACS cases (80% training / 20% test split) from 2014 to 2021.

NYC Algorithmic Tools Compliance Report — Family Team Conference Model (FTC)

What it does

Deciding (Analytical AI)

Human decides

The model produces a monthly rank-ordered list of prevention cases by predicted risk score. Senior Division of Prevention Services staff review this list and decide which families to assign ACS Child and Family Specialists to for the FTC Reassessment Planning Conference. No automated action is taken directly from model output.

NYC Algorithmic Tools Compliance Report — Family Team Conference Model (FTC)

Outputs

A recommendation or prediction

Anonymized data

A monthly rank-ordered list of open prevention cases involving children whose families have the highest likelihood of being in an indicated SCR investigation within 24 months from date of observation. The list is advisory — it surfaces cases for senior staff review and does not itself constitute a decision about any individual family.

NYC Algorithmic Tools Compliance Report — Family Team Conference Model (FTC)

Run by

Administration for Children's Services (ACS)

ACS, specifically its Division of Policy, Planning, and Analysis and Division of Prevention Services, deploys and operates this model to support timely, data-informed decision-making for families engaged in child welfare prevention services.

NYC Algorithmic Tools Compliance Report — Family Team Conference Model (FTC)

Built by

Not stated by the NYC Office of Technology and Innovation.

Kept for

Retained Not specified in the register

The register does not specify the retention period for model outputs (the monthly ranked list) or for the underlying ACS administrative data used as input. Training data covered closed investigations from 2014 to 2021.
Duration: Not specified in the register

Shared with

Available to the accountable organization

The monthly ranked list of prevention cases is available to senior staff in ACS's Division of Prevention Services and to ACS Child and Family Specialists assigned to FTC Reassessment Planning Conferences.

Not available to me

The ranked risk score list is an internal operational tool used by ACS staff. Individual families do not receive their risk score or their placement on the ranked list. The register states no identifying information is collected or disclosed during prediction.

Stored

Not stated by the NYC Office of Technology and Innovation.

How to read the colours

Can it identify you?

Anonymized data: Data about people with the link to who is broken. Stripped of identifiers, blurred, aggregated, or noised so this system can’t reasonably tie a record back to an individual.
Pseudonymous data: Each person’s data is tied to a token (hash, ID, template) that lets this system recognise the same person across events, but the token itself doesn’t reveal a name. Reidentification is possible with extra information.
Identifiable data: The data either contains a direct identifier (name, address, account name, recognisable face or voice, plate number) or carries a token this system uses to look up legal identity during processing.

Who completes the loop?

Human decides: This mode suggests; a person decides what to do next. The AI is always advisory — a human is in the loop on every decision. Example: a triage tool ranks cases for a clinician who chooses which to see first.
Human executes: This mode decides; a person carries out the result. Example: an optimizer plans the day’s trash-collection routes, and drivers run them.
Autonomous: This mode decides and acts on its own. No person reviews each decision or carries out the resulting action.

Definitions from the DTPR standard. Amber is about your data, violet about who decides. The fuller the shape and the deeper the colour, the more identifying the data or the less a person is involved.

What you can do

Ask about this system

Questions go to the NYC Office of Technology and Innovation, not the vendor.

Your rights

Right to Be Informed of AI UseFamilies engaged in ACS prevention services should be informed that a predictive risk model is used to prioritize cases for enhanced attention at the FTC Reassessment Planning Conference. The register does not specify whether families are directly notified that their case has been scored or ranked. Contact ACS for information about how the FTC model affects your case.
Right to a Human ReviewAll decisions flowing from the FTC model output are made by human staff. Senior Division of Prevention Services staff review the ranked list and determine case assignments. The FTC Reassessment Planning Conference itself is a structured human-facilitated meeting that includes the family, ACS Child and Family Specialists, and provider agencies. No automated decision is made solely on the basis of the model's risk score.
Right to Algorithmic TransparencyACS has disclosed the existence of the FTC model in the NYC Algorithmic Tools Compliance Report, including its purpose, input data, training data, and the supervised human review process. The report is publicly available at the NYC Open Data portal. Individuals seeking further detail about how the model works may contact ACS's Division of Policy, Planning, and Analysis.

Risks and safeguards

Civil liberties harmPredictive risk scoring in child welfare contexts can have chilling effects on families' willingness to engage with services, and may entrench surveillance of already-marginalized communities. Risk scores could influence resource allocation in ways that affect family autonomy and child welfare outcomes without transparent recourse.Safeguard: The model is supervised — senior staff review ranked output and make all actual decisions. The register states no identifying information is collected or disclosed during prediction. The FTC conference structure is designed to include family voice and participation. The model informs conference assignment, not case closure or removal decisions.
Reputational harmFamilies assigned elevated risk scores may be stigmatized or subjected to heightened scrutiny based on algorithmic labeling rather than direct caseworker assessment, potentially affecting how ACS staff and provider agencies perceive and interact with them.Safeguard: The output is a ranked advisory list reviewed by senior staff, not a label affixed to a family record. The register notes the model supports FTC conference prioritization — not a final determination — and the conference format is explicitly designed to incorporate family voice and provider collaboration.