Our projects

Our research projects are all about data analytics for social impact, with most projects focusing on impact evaluation, predictive risk modelling or wellbeing surveys.  Projects are listed from most recent to least recent.

Allegheny Family Screening Tool

Location: Allegheny County – PA, United States
Partner/s: Children’s Data Network (University of Southern California), Carnegie Mellon University, Allegheny County Department of Human Services
Timeframe: 2014-ongoing

Dedicated project page

Researchers led by Rhema Vaithianathan modelled, designed and supported implementation of this world-first child welfare predictive analytics tool.  The Allegheny Family Screening Tool (AFST) uses rich administrative data to generate a screening score for incoming calls alleging child maltreatment and neglect. The score is an additional piece of information that helps call screeners as they decide whether to open an investigation.  Allegheny County introduced this decision support tool with the aim of improving accuracy and consistency of call screening decisions. An independent impact evaluation of the tool was completed by researchers at Stanford University in March 2019.  The evaluators’ findings included that use of the tool improved the accurate identification of children in need of services and was associated with a modest reduction in racial disparities in case openings.

Cuccaro-Alamin, S., Foust, R., Vaithianathan, R., & Putnam-Hornstein, E. (2017). Risk assessment and decision making in child protective services: Predictive risk modeling in context. Children and Youth Services Review, 79, 291-298.

The authors, including Rhema Vaithianathan, review the literature and provide a context for predictive risk modelling in the current risk assessment paradigm in child protective services. They describe how predictive analytics or predictive risk modelling using linked administrative data may provide a useful complement to current approaches.

Vaithianathan, R., Jiang, N., Maloney, T., Nand, P., & Putnam-Hornstein, E. (2017). Developing predictive risk models to support child maltreatment hotline screening decisions: Allegheny County methodology and implementation. Auckland: Centre for Social Data Analytics.

This methodology document traces the development and implementation of the Allegheny Family Screening Tool (AFST), a child welfare predictive risk-modelling tool built by a research team led by Rhema Vaithianathan. The AFST is intended to support call screening of child maltreatment referrals by Allegheny County Department of Human Services.  The methodology document includes an overview of existing practice, methodology and performance metrics of the model and details of the external validation and next steps for the model rebuild.

Chouldechova, A., Putnam-Hornstein, E., Benavides-Prado, D., Fialko, O., & Vaithianathan, R. (2018). A case study of algorithm-assisted decision making in child maltreatment hotline screening decisions. Proceedings of Machine Learning Research, 1-15.

Diana Bendavides-Prado, Rhema Vaithianathan, Oleksandr Fialko (CSDA research fellow 2017-2018), Alexandra Chouldechova and Emily Putnam-Hornstein describe their work on developing, validating, fairness auditing, and deploying a risk prediction model in Allegheny County, PA, USA. They discuss the results of their analysis to-date and highlight critical problems and data bias issues that present challenges for model evaluation and deployment.

Vaithianathan, R., Kulick, E,  Putnam-Hornstein, E & Benavides Prado, D. (2019). Allegheny Family Screening Tool:  Methodology, Version 2. Centre for Social Data Analytics. Auckland: Centre for Social Data Analytics.

This methodology report describes changes to the Allegheny Family Screening Tool (AFST), building upon and updating the original methodology report, Developing Predictive Risk Models to Support Child Maltreatment Hotline Screening Decisions (March 2017). Modifications implemented include changes to specific predictor fields used in the model itself, the modelling methodology, and County policies concerning the tool’s use,upholding Allegheny County’s ongoing commitment to transparency by continuing to inform the community about changes to the tool and the County’s policies.

Vaithianathan R, Putnam-Hornstein E, Chouldechova A, Benavides-Prado D, Berger R. Hospital Injury Encounters of Children Identified by a Predictive Risk Model for Screening Child Maltreatment Referrals: Evidence From the Allegheny Family Screening Tool. JAMA Pediatr. Published online August 03, 2020. doi:10.1001/jamapediatrics.2020.2770

This paper confirms that children identified as at risk by the Allegheny Family Screening Tool (AFST), a predictive risk modelling tool that supports child protection decisions in Allegheny County (PA, United States), are also at considerably heightened risk of hospitalisation injury.  The authors, including Rhema Vaithianathan and Diana Benavides Prado, used a one-off linkage between the child protection system and Pittsburgh Children’s Hospital data to show that children identified as at risk by the AFST are indeed at considerably heightened risk of injury, abuse, and self-harm hospitalisation.  The authors also identified a sharply heightened hospitalisation risk for children receiving the highest AFST scores and found that the AFST is particularly sensitive to the risk facing White children, which means the child protection system is potentially under protecting those children.

Key AFST documents including Methodology and Methodology V2, Ethics Report, Frequently Asked Questions and Impact Evaluation (March 2019) along with press coverage are available on the Allegheny Analytics site.

Douglas County Decision Aide

Location: Douglas County - CO, United States
Partner/s: Douglas County Department of Human Services, Children’s Data Network (University of Southern California)
Timeframe: 2017-2020

Dedicated project page

In early 2017 researchers developed a prototype child welfare (maltreatment) predictive risk model for Douglas County, Colorado.  The Douglas County Decision Aide is a decision support tool designed to help triage incoming calls alleging child maltreatment and neglect.  The tool uses data from child welfare and public welfare eligibility systems.  The DCDA was implemented by Douglas County leadership as a year-long randomised controlled trial in February 2019, with results on the impact of the tool expected in mid-2020.

Cuccaro-Alamin, S., Foust, R., Vaithianathan, R., & Putnam-Hornstein, E. (2017). Risk assessment and decision making in child protective services: Predictive risk modeling in context. Children and Youth Services Review, 79, 291-298.

The authors, including Rhema Vaithianathan, review the literature and provide a context for predictive risk modelling in the current risk assessment paradigm in child protective services. They describe how predictive analytics or predictive risk modelling using linked administrative data may provide a useful complement to current approaches.

Vaithianathan, R., Jiang, N., Maloney, T., Nand, P., & Putnam-Hornstein, E. (2017). Develping predictive risk models to support child maltreatment hotline screening decisions: Allegheny County methodology and implementation. Auckland: Centre for Social Data Analytics.

This methodology document traces the development and implementation of the Allegheny Family Screening Tool (AFST), a child welfare predictive risk-modelling tool built by a research team led by Rhema Vaithianathan. The AFST is intended to support call screening of child maltreatment referrals by Allegheny County Department of Human Services.  The methodology document includes an overview of existing practice, methodology and performance metrics of the model and details of the external validation and next steps for the model rebuild.

Chouldechova, A., Putnam-Hornstein, E., Benavides-Prado, D., Fialko, O., & Vaithianathan, R. (2018, January). A case study of algorithm-assisted decision making in child maltreatment hotline screening decisions. In Conference on Fairness, Accountability and Transparency (pp. 134-148).

Diana Bendavides-Prado, Rhema Vaithianathan, Oleksandr Fialko (CSDA research fellow 2017-2018), Alexandra Chouldechova and Emily Putnam-Hornstein describe their work on developing, validating, fairness auditing, and deploying a risk prediction model in Allegheny County, PA, USA. They discuss the results of their analysis to-date and highlight critical problems and data bias issues that present challenges for model evaluation and deployment.

Vaithianathan, R., Dinh, H., Kalisher, A., Kithulgoda, C. I., Kulick, E., Mayur, M., … Putnam-Hornstein, E. (December 2019). Implementing a Child Welfare Decision Aide in Douglas County: Methodology Report. Centre for Social Data Analytics.

This report summarises the development and implementation of the Douglas County Decision Aide (DCDA), a child welfare predictive risk-modelling tool built and deployed by the Centre for Social Data Analytics. The DCDA is intended to support call screening of child maltreatment referrals made to the Douglas County Department of Human Services (CO, United States). The methodology outlines how Douglas County was triaging referrals prior to the deployment of the DCDA and how the DCDA has changed that, as well as details of how the DCDA tool was built and validated and how ethical concerns were identified and responded to.

Hello Baby Program - Proactive Child Welfare Predictive Risk Model

Location: Allegheny County – PA, United States
Partner/s: Allegheny County Department of Human Services, Children’s Data Network (University of Southern California)
Timeframe: 2018-

Dedicated project page

Allegheny County’s Hello Baby Program is intended to provide every family of a new-born universal access to information and resources and differentiated and intensive support for families with complex challenges and needs. The CSDA research team was engaged by Allegheny County Department of Human Services to develop a screening tool that would allow for identification of families with highest needs.  The agency will contract externally for programmatic intervention and independent evaluation.

Vaithianathan, R., Maloney, T., Putnam-Hornstein, E., & Jiang, N. (2013). Children in the public benefit system at risk of maltreatment: Identification via predictive modeling. American Journal of Preventive Medicine, 45(3), 354-359.

This paper co-authored by Rhema Vaithianathan and Tim Maloney (CSDA co-director 2016-2019) outlines the method of using an automated predictive risk model to identify children at high risk of maltreatment, as well as discussing the strategic targeting of prevention activities toward these individuals.

Vaithianathan, R., Rouland, B., & Putnam-Hornstein, E. (2018). Injury and mortality among children identified as at high risk of maltreatment. Pediatrics, 141(2), e20172882.

This study by Rhema Vaithianathan, Bénédicte Rouland (CSDA research fellow 2016-2018) and Emily Putnam-Hornstein explored a model which assigned risk scores for the risk of a substantiated finding of maltreatment to children born in New Zealand in 2010, to see whether children at the highest 10% and 20% of risk would have an elevated chance of injury or death in early childhood. The study found that children assessed as being “very high risk” (highest 10%) and “high-risk” (highest 20%) had 4.8 times and 4.2 times respectively higher post-neonatal mortality rates than other children.

Vaithianathan, Rhema; Diana Benavides-Prado and Emily Putnam-Hornstein. Implementing the Hello Baby Prevention Program in Allegheny County. Centre for Social Data Analytics. Auckland, New Zealand. September 2020.

This report summarises the development and implementation of the Hello Baby program which is supported by a predictive risk modelling (PRM) tool, built and deployed by the Centre for Social Data Analytics. Hello Baby is a new universal tiered program designed to prevent harm and protect children, introduced as a pilot program by the Allegheny County Department of Human Services in September 2020.  The County will use the PRM tool to establish eligibility for higher intensity services, alongside self-referral and professional referral. The methodology provides background to the introduction of Hello Baby, explains the selection of the PRM tool as an eligibility pathway, describes how historical data was used to train and validate the PRM tool and explains how the tool will be used as part of the implemented Hello Baby program. Additional documentation about the Hello Baby program including ethics reports and FAQs is found on the Allegheny County website.

MyDay Survey with Health Education England, Working Across Wessex

Location: Wessex, England
Partner/s: Health Education England, Working Across Wessex, University of Southampton, City, University of London, University of Zurich
Timeframe: 2017-2018

Dedicated project page

Researchers used MyDay, a purpose-built web-based survey tool, to collect anonymous data from up to 2100 junior doctors about their wellbeing at work.  Participants listed tasks they did in five short windows of time, and how they felt doing each task.  The research set out to provide a rich and useful picture of the experienced wellbeing of this key workforce.

Hockey, P., Vaithianathan, R., Baeker, A., Beer, F., Goodall, A. H., Hammerton, M., Jarvis, R., Brock, S., & Lorimer, L. (2020). Measuring the working experience of doctors in training. Future Healthcare Journal, 7(3), e17–e22. https://doi.org/10.7861/fhj.2020-0005

The researchers, including CSDA’s Rhema Vaithianathan and Larissa Lorimer, use an online tool (‘MyDay’) to look for an association between tasks and emotional affect for 565 doctors in training. They find that the participant trainee doctors spent a quarter of their time at work on paperwork or clinical work that did not involve patients, which was associated with more negative emotions.  Positive emotions were associated with breaks, staff meetings, research, learning and clinical tasks that involved patients.  Trainee doctors reporting that they had considered leaving medicine reported more negative feelings. The authors conclude that systematic workplace changes, like regular breaks, reduced paperwork and improved IT systems could contribute to positive workplace experiences and reduce the intention to quit among doctors in training.

MyDay Survey with Warren and Mahoney

Location: Auckland, New Zealand
Partner/s: Warren and Mahoney
Timeframe: 2018

Dedicated project page

CSDA researchers partnered with an Australasian architectural practice to explore the feasibility of combining a survey on workplace wellbeing with location data, to understand the impact of physical spaces on wellbeing at work.

An Ethical Framework for Social Policy Applications of Predictive Analytics

Location: Auckland, New Zealand
Partner/s: University of Auckland (NZ)
Timeframe: 2018-2020

Dedicated project page

The objective of this research is to identify, analyse, and respond to ethical issues generated by social policy uses of predictive analytics.   Outputs will include a set of practical guidelines for the ethical use of predictive analytics in social policy contexts, based on the researchers’ experience in policy applications of predictive analytics and detailed research including consultation with policy makers.  This research is funded by a 2017 NZ Royal Society Marsden Fund Grant.  Rhema Vaithianathan is an associate investigator on this project - the principal investigator is Professor Tim Dare (University of Auckland, New Zealand).

Homelessness Predictive Risk Model for Allegheny County, US

Location: Allegheny County, PA, United States
Partner/s: Allegheny County Department of Human Services
Timeframe: 2017-

Dedicated project page

CSDA researchers used rich administrative data to build a predictive risk modelling tool to support the prioritisation of clients for longer term housing programs in Allegheny County, Pennsylvania (United States). The Allegheny Housing Assessment (AHA) was deployed by Allegheny County in September 2020. The CSDA research team continues to work with the County and partners to fine tune the model and the County is planning to issue a solicitation for an independent evaluation of the AHA in 2021.

Documentation for the AHA includes the Methodology, Frequently Asked Questions, Report on Client Focus Groups, Independent Ethical/Data Science Review and Allegheny County’s Response to the Ethical/Data Science Review.

  • The Allegheny Homelessness Tool (August 2019) - a poster presented at the Bloomberg Data for Good Exchange (D4GX) in New York, September 2019.
  • Using Predictive Risk Modeling to Prioritize Services for People Experiencing Homelessness in Allegheny County. Centre for Social Data Analytics. Auckland, New Zealand. September 2020.
    This report summarises the development and implementation of the Allegheny Housing Assessment, a predictive risk modelling tool, built and deployed by the Centre for Social Data Analytics for the Allegheny County Department of Human Services, Pennsylvania, United States. The AHA will be used to support the prioritisation of people for longer term housing programs. A prioritisation tool is needed because the demand for housing exceeds supply. The AHA replaces the VI-SPDAT, a prioritisation system without local validation. The methodology provides background to the development of the AHA including a comparison between the AHA and the previously used VI-SPDAT. The methodology provides background on homeless services in Allegheny County and the decision to explore the use of a predictive risk modelling tool, before discussing elements including community engagement, the methodology used to develop the tool, accuracy and the business process it will be used within.

Understanding the factors influencing wellbeing among Indigenous youth in the Northern Territory

Location: Australia
Partner/s: University of Sydney, Melbourne Institute of Applied Economic and Social Research, University of Western Australia, Menzies School of Health Research, University of Melbourne, Danila Dilba Health Service (Darwin), Centre for Remote Health, Massey University (NZ)
Timeframe: 2018-2023

Dedicated project page

This mixed-methods project studies the factors that influence social and emotional wellbeing (SEWB) of Indigenous youth in the Northern Territory. It uses administrative data and econometric techniques to identify communities where youth have higher or lower SEWB than expected. These locations are then visited to ask Indigenous youth about their perspective on SEWB (giving them a voice). Research-to-practice translation will focus on cultural safety, respect and benefits for Indigenous youth. Rhema Vaithianathan is an associate investigator on this project.  The chief investigator is Associate Professor Stephanie Schurer (University of Sydney).  This five-year project is funded via the National Health and Medical Research Council (NHMRC) Targeted Call for Research.

Our publications

Read our latest reports and academic papers.

List of publications