Advancing APS Data Integrity: From Error Reduction to Reliable Insights
 

By Karl Urban, Senior Research Manager, APS TARC

December 2025

You may have noticed more emphasis this year from the APS TARC on data quality. For us, this means doing what we can to help adult protective services (APS) programs improve the quality of the data they report to the National Adult Maltreatment Reporting System (NAMRS).  
 

Here are five reasons why this is important for APS programs: 
 

1. We can’t escape the coming use of artificial intelligence (AI) in APS. Right now, there is not much use of AI in APS, but that is going to change. The tools are becoming much more sophisticated, and they have the potential to improve casework and decision-making and reduce workload. At the APS TARC, we added AI to our ongoing list of topics for potential best practice/innovations analysis and technical assistance; however, we didn’t recommend AI as a topic for this year because it still feels premature. With only a couple of exceptions, AI is not built into APS supports systems yet. However, based on your recent inquiries and recommendations for TA topics, we all know it is coming. The question is will we be ready for it?   

One thing that will define readiness is the quality of data in the APS system. At its core, AI is built on data processing. We have all heard of and probably have started using “large language models” (e.g., ChatGPT, Claude, Gemini). I recently heard an AI expert on a podcast lamenting the fact that large language models are, well, large. That is, they are trying to use the entire universe of available information (literally) in their algorithms when they should be using targeted data sets with reliable information. One consequence of this is what are called “hallucinations” — that is, inaccurate results because the quality of information was not good. The future of effective use of AI depends on turning reliable data into insights, while actively mitigating risks like hallucinations.
 

2. Improving quality assurance (QA) programs is a priority for APS programs. Within the past few years, multiple APS programs have implemented initiatives to create or upgrade their QA programs. As we have described in the past APS TARC TA materials (here, here, and here), QA must be built on a foundation of quality data. One of the most promising use cases for AI is to support quality assurance by uncovering patterns and trends in casework—whether they indicate potential issues, opportunities for improvement, or best practices. For example, we worked with ACL on a use case in which a machine learning algorithm predicted substantiation based on case details. While the substantiation decision may be correct, the discordance between the case documentation/data and the predicted outcome could be used to flag cases for QA review (or supervisor review before closure).
 

3. Improving case management systems is a priority for APS programs. Within the past few years, multiple APS programs have invested in improved systems to generate more and better data. Programs want more and better data because of the next reason. 
 

4. Improving the use of data as a management tool is a priority for APS programs. We have provided general TA webinars (here, here, here, and here) related to how to use data and have provided state-specific workshops (and are always available for further consultation) on how to use data for successful program management. As programs increasingly use data for decision-making, it needs to be complete and accurate.
 

5. Improving the use of APS administrative data is a priority for NAPSA. The NAPSA BEACON (Benchmarking Elder Abuse Compared to Other Needs) project is using APS administrative data to help educate policymakers about the scope of elder mistreatment as investigated by APS programs. The goal is to provide credible data at the local level and put results in context by comparing elder mistreatment prevalence to other older adult concerns. To do so requires us to collectively continue to improve data quality and think about the best ways to use administrative data.
 

In summary, poor data quality—whether originating from human input, system errors, or AI-generated outputs—can lead to flawed decisions and undermine APS’s credibility.  This means that the data we use must be accurate and reliable. To this end, we are focusing TA efforts this year, beyond our regular NAMRS support, on improving data quality. Please check out the webinars and other resources as they are released.  
 

We look forward to continuing to work with you to achieve these goals and prepare for what is to come. 
 

The APS Blog is updated regularly with posts from contributing authors and new publications from the APS TARC.

How did we do? Take our quick customer satisfaction survey to give us feedback.