SECTION 1. Short title.

SEC. 2. Findings.

(1) 4 to 8 children in the United States die every day from abuse or neglect at the hands of their parents or caretakers;

(2) according to the 2016 Child Maltreatment report of the Department of Health and Human Services, there were 1,700 fatalities resulting from child maltreatment reported in fiscal year 2016, an increase of 7 percent from the previous year;

(3) in 2016, approximately 676,000 children were victims of abuse and neglect;

(4) predictive analytics can be used as a tool in helping to identify the children most at risk of maltreatment;

(5) harnessing and analyzing administrative data can help to drive better outcomes for children and families, as well as assist child protective service workers in delivering much needed services to children who are most at risk; and

(6) we should explore every possible opportunity to do better for our children, and predictive analytics tools provide such an opportunity, if such tools are implemented with appropriate oversights and safeguards.

SEC. 3. Funding for the development of a predictive analytics pilot program to help children and families who come to the attention of the child welfare system.

(1) by striking “The Secretary” and inserting the following:

“(A) IN GENERAL.—The Secretary”;

(2) by striking “of research-based strategies for risk and safety assessments relating to child abuse and neglect” and inserting “of—

“(i) research-based strategies for risk and safety assessments relating to child abuse and neglect; and

“(ii) the predictive analytics pilot program described in subparagraph (B).”; and

(3) by adding at the end the following:

“(B) PREDICTIVE ANALYTICS PILOT PROGRAM.—

“(i) GRANTS FOR PREDICTIVE ANALYTICS PILOT PROGRAM.—The Secretary may award grants under this subparagraph to eligible entities to develop predictive analytics programs that provide for the development of research-based strategies for risk and safety assessments relating to child abuse and neglect, for the purpose of helping children and families who come to the attention of the child welfare system.

“(ii) ELIGIBLE ENTITIES.—To be eligible to receive a grant under this subparagraph, an entity shall be a State, local governmental entity, or Indian tribe.

“(iii) APPLICATION REQUIREMENTS.—An eligible entity (or partnership of entities) that seeks funding under this subparagraph shall submit to the Secretary—

“(I) a description of the predictive analytics pilot;

“(II) a description of the geographic area to be served;

“(III) the expected outcomes of the pilot;

“(IV) a description of the data intended for use, including how the standards for the predictive analytics model are valid, reliable, equitable, and useful;

“(V) a description of the extent to which the eligible entity has access to sufficient, available quality datasets to build and maintain an accurate predictive model;

“(VI) the technical resources, including staff training, needed to carry out the pilot;

“(VII) a description of how the predictive analytics tool will fit within and be complementary to the strategy of child protection work in the geographic area that the model serves;

“(VIII) a description of how the entity believes the use of predictive analytics will—

“(aa) assist the entity (or partnership or entities) in better identifying and providing appropriate services to children most at risk of abuse and neglect;

“(bb) reduce bias based on race, sex, religion, national origin, age, and disability; and

“(cc) prevent any disparate treatment of children and families based on such characteristics;

“(IX) the procedures to monitor and prevent unintended bias in child protective practices as a result of the use of predictive analytics, including procedures on who has access to the tool, who intends to use the predictive model and its assessment, and how to ensure that the tool will not be used as the sole decisionmaker in determining risks to children;

“(X) a description of how the agency solicited and considered input from and collaborated with members of the community, child welfare organizations, relevant government agencies, privacy advocates, civil rights advocates, families who have previously had contact with the child welfare system, and relevant community-based or faith-based organizations in developing, testing, and amending the model;

“(XI) assurances that there will be an open and transparent process to the public in the development and implementation of the program, including—

“(aa) the development of the predictive analytics model;

“(bb) the determination of what data will be used and the relative importance of each data point used to determine risk; and

“(cc) the implications and limitations of the pilot program;

“(XII) assurance the pilot will comply with all privacy and civil rights laws; and

“(XIII) such other information as the Secretary may require.

“(iv) AWARDING GRANTS.—The Secretary shall select up to 5 eligible entities (or partnership of entities) to participate in the pilot program under this subparagraph. In awarding grants under this subparagraph, the Secretary shall prioritize eligible entities that have not yet implemented a predictive analytics system for the purpose described in clause (i).

“(v) TECHNICAL ASSISTANCE, EVALUATIONS.—

“(I) IN GENERAL.—The Secretary shall allocate up to 30 percent of amount reserved under clause (vi) for the purposes of evaluating pilots and providing technical assistance to eligible entities as follows:

“(aa) The eligible entity (or partnership of entities) may use amounts provided under this subclause to enter into an agreement with a third party to provide technical assistance to grantees on the development, design, and testing of predictive analytic model of each eligible entity, including compliance with all applicable privacy laws.

“(bb) The eligible entity (or partnership of entities) may use amounts provided under this subclause to enter into an agreement with a third-party, independent evaluator with expertise in predictive analytics in the social sciences to evaluate each pilot program to determine the impact of such program, consistent with subclauses (II) through (V).

“(II) EVALUATION SCOPE.—Each evaluation under subclause (I)(bb) shall measure the impact of the pilot program on—

“(aa) the ability to accurately predict which children are most at risk of further abuse or neglect;

“(bb) the ability of the entity to more effectively target services to the highest risk families;

“(cc) rates of child abuse and neglect;

“(dd) child fatalities;

“(ee) child removal and reunification rates;

“(ff) reducing bias based on race, sex, religion, national origin, age, and disability;

“(gg) preventing any disparate treatment of children and families based on such characteristics;

“(hh) other measures of child and family well-being; and

“(ii) any other measures deemed appropriate by the independent third-party evaluator in consultation with the eligible entity.

“(III) METHODOLOGIES TO BE USED.—The evaluation under subclause (I)(bb) shall use experimental designs using random assignment or other reliable, evidence-based research methodologies that allow for the strongest possible causal inferences when random assignment is not feasible.

“(IV) INTERIM REPORT.—Not later than 1 year after the pilot program has commenced, the independent evaluator described in subclause (I)(bb) shall submit to the Secretary, the Committee on Health, Education, Labor, and Pensions of the Senate, and the Committee on Education and the Workforce of the House of Representatives an interim report on the progress of such program.

“(V) FINAL REPORT.—Not later than 6 months after completion of the pilot program, the independent evaluator described in subclause (I)(bb) shall submit to the Secretary, the Committee on Health, Education, Labor, and Pensions of the Senate, and the Committee on Education and the Workforce of the House of Representatives a written report that includes the results of the evaluation and the conclusion of the evaluator as to whether the pilot program had a positive impact on children and families that come to the attention of the child welfare system of the applicable State, together with information on the unique factors that contributed to the success or failure of the pilot program, the challenges faced in attempting the pilot program, and information on the improved future delivery of this or similar predictive analytics pilot programs or other programs.

“(vi) RESERVATION OF FUNDS.—Of the amounts made available under section 112 for fiscal year 2018, the Secretary shall reserve $10,000,000 for purposes of carrying out this subparagraph, and the amount so reserved shall remain available through the end of fiscal year 2022.

“(vii) SUNSET.—The authority under this subparagraph shall cease to have force or effect on the day after the final report required under clause (v)(V) is submitted to the Secretary, the Committee on Health, Education, Labor, and Pensions of the Senate, and the Committee on Education and the Workforce of the House of Representatives.”.