Hurricane Score

Overview

Development stage

DeployedSource: c) Deployed – The use case is being actively authorized or utilized to support the functions or mission of an agency.

High-impact designation

High-impactSource: a) High-impact

Impact justification

Not reported

Start date

2019-02-01 (day precision)Source: 2019-02-01T00:00:00

Withheld from public reporting

Not reported

Mission

Topic area

Law Enforcement

Operational functions

Claims and case processing Derived by keyword rules; see methodology

Problem the AI is intended to solve

This use case intends to solve the problem of understanding which noncitizens in Enforcement and Removal Operations’ Alternatives to Detention - Intensive Supervision Appearance Program are most likely to abscond.

Expected benefits

The Hurricane Score helps officers quickly evaluate substantial amounts of case information across thousands of Alternatives to Detention - Intensive Supervision Appearance Program participants. By surfacing a risk indicator based on observed absconding patterns, it can provide additional insight that might not be apparent from manual review alone. This supports more consistent and efficient case reviews and helps officers allocate case management resources more effectively while maintaining individualized assessments.

System outputs

Once individuals are enrolled in Enforcement and Removal Operations’ Alternatives to Detention - Intensive Supervision Appearance Program (ATD-ISAP), officers periodically review each case to determine whether the current level of case management and technology assignment remains appropriate or should be adjusted. During case reviews, an analyst or officer provides the Hurricane Score model with information already known about an ATD-ISAP participant, including case management details and participant actions. The model is a quasi-binomial, binary classification machine learning (ML) model trained on inactive ATD-ISAP case data to identify patterns associated with prior absconding behavior. Based on the provided inputs, the model outputs a score from 1 to 5, with higher scores indicating a higher model-estimated risk that the individual may abscond. Officers may then consider this score, along with many other factors, when determining whether current levels of case management or technology assignment remain appropriate or should be adjusted.

Technology

AI classification

Classical / predictive MLSource: Classical/Predictive Machine Learning: Models trained on data to make predictions or classifications based on identified patterns or relatio…

System name(s)

Not reported

Custom-developed code

Yes

Public source code

Not reported

Sourcing

How it was built

Developed in-houseSource: b) Developed in-house

Vendor (as reported)

Not reported

Vendors (standardized)

None on the standard list

A vendor is the supplier named by the agency. It does not identify the underlying model or AI technology.

Data and privacy

Involves PII

Yes

Privacy Impact Assessment

www.dhs.gov ↗

Authorization to Operate

No

Demographic features

b) Sex; c) Age

Training and evaluation data

Inactive case data from individuals enrolled in the ATD-ISAP program.

Federal Data Catalog

Not reported

Governance

8 of 8 minimum-practice questions answered. A blank answer means the agency reported nothing; it does not mean the practice is absent.

Potential impacts and how they were identified

Predictive ML techniques can produce misleading results, such as false positives, which could impact case management decisions if relied upon as a primary factor. For instance, an inaccurate hurricane score might lead to stricter or more lenient compliance or technology requirements for an individual. ERO mitigates this by using the score as one of many factors in determining case management or technology levels for individuals in the ATD program.