Project summary
The University of Wollongong in partnership with QPS Benchmarking and Warrigal Care has been awarded an ARIIA grant for their project ‘Implementing an AI Tool NutriTrack to Enhance Nutrition Care in Aged Care’.
Malnutrition has severe health consequences for older people, e.g., weakened immune system, damaged organs, and increased mortality. The methods for screening malnutrition in aged care are time-consuming and thus not timely conducted. Medical and aged care staff spend much time capturing valuable client data, including diagnosis, assessment, and malnutrition indicators, in electronic health records (EHR), often in free-text care notes.
Recent advancements in artificial intelligence and natural language processing provide great opportunities for extracting indicator data from EHR and automating clinical diagnoses and risk predictions [1, 2].
We seek funding to expand our preliminary malnutrition prediction model that has been trained on EHR data from 4,405 clients in 40 residential care facilities [3] to the national dataset captured by QPS Benchmarking. We will scale up and commercialise the decision support tool NutriTrack that we will develop. We will integrate NutriTrack into the quality improvement system to improve nutrition care governance, care model and workflow. This will bring in further benefits for dementia, social isolation, mental health and wellbeing, and palliative care.
Will conduct pre- and post-implementation in the partner aged care organisation Warrigal Care to evaluate the success of NutriTrack and promote the innovative translational research project.
Project outcomes
Background and Aims
Malnutrition can severely impact older people, weakening their immune system, damaging organs, increasing hospitalisation rates, and raising mortality risks. Existing manual processes for detecting malnutrition in aged care are slow and often delayed. Aged care staff spend a lot of time recording health assessments, weight, and fluid intake in electronic health records (EHR), which presents a unique opportunity to automate nutrition assessments. The NutriTrack project aimed to develop and integrate an AI-driven tool into ValleyView Residence’s EHR system to improve the quality of nutrition care for older people. By automating assessments, NutriTrack ensures timely and accurate malnutrition detection, leading to better health outcomes and improved care for older adults.
What We Did
We explored how artificial intelligence (AI) technology could extract valuable insights about older people from nurses’ daily care notes in electronic health records (EHR). We developed advanced computer programs to analyse these records, focusing on nutrition and hydration data. Our research findings were shared with the scientific community through a published paper in a leading international journal, with additional papers under development. The project validated the evidence-based nutrition practices already embedded in ValleyView’s PCS system, such as the Malnutrition Universal Screnint Tool (MUST), the International Dysphagia Diet Standardisation Initiative (IDDSI) framework, which provides evidence-based guidelines, for texture modification, to systematically tailor food and drink preparation to meet the individualised swallowing and nutritional needs of residents with dysphagia. PCS also allows monitor daily fluid intake.
Outcomes
The project delivered both research and practical outcomes:
Research outcomes: The NutriTrack project led to the publication of a peer-reviewed journal article, with further papers in progress. These contributions advance global understanding of AI applications in aged care.
Practical outcomes: The project developed advanced AI tools for malnutrition risk identification, offering a proactive approach to improving resident care. By validating and reinforcing best-practice nutrition care at the partner organisation, ValleyView Residence, the project demonstrated how person-centred nutrition care can be effectively integrated with a robust clinical information system like PCS. This synergy enhances the delivery of nutrition care by leveraging PCS’s existing capabilities, including the MUST tool, the IDDSI framework, and daily fluid intake monitoring. Furthermore, the project optimised the application of these PCS functions, ensuring more accurate, timely, and actionable insights into residents’ nutritional and hydation needs. This not only strengthens governance and clinical decision-making but also sets a new benchmark for evidence-based nutrition care in aged care settings.
Impact on Aged Care and Workforce
The NutriTrack project strengthened evidence-based practices, such as MUST and IDDSI frameworks, embedded in ValleyView’s PCS system. By validating a technology-supported nutrition care model, this project enhances aged care managers’ confidence in digital health systems, encourages aged care staff to use digital tools to deliver timely, accurate, and proactive nutrition care, which benefits the health and well-being of older adults. The project also demonstrates that a well-designed clinical information system can support better governance and decision-making in aged care, promoting digitally innovative and evidence-based nutrition care in the aged care sector.
Resources
Peer-reviewed scientific paper:
D. Vithanage, C. Deng, L. Wang, M. Yin, M. Alkhalaf, Z. Zhang, Y. Zhu, P. Yu, “Adapting Generative Large Language Models for Information Extraction from Unstructured Electronic Health Records in Residential Aged Care: A Comparative Analysis of Training Approaches”, Journal of Healthcare Informatics Research, 2025 (impact factor: 5.4, Q1 journal). Link: https://link.springer.com/article/10.1007/s41666-025-00190-z
Computer program:
NutriTrack malnutrition calculator. We are building the web site to host it, which will be made available by the end of July.
Next Steps
- Continue the collaboration with ValleyView Residence and PCS Software to integrate the computer programs we developed into the commercial electronic clinical information system for aged care - PCS.
- Publish the second research article, which has been revised and is pending journal editorial decision: D. Vithanage, P. Yu, Q. Xie, H. Xu, L. Wang, C. Deng (revision submitted) A Comprehensive Evaluation of Large Language Models for Information Extraction from Unstructured Electronic Health Records in Real-World Clinical Settings, Computers in Biology and Medicine (impact factor: 7, Q1 journal).
- Finalising and submitting the manuscript to the journal Computers in Biology and Medicine (impact factor: 7, Q1 journal). Q. Duong, D. Vithanage, C. Deng, J. Chow, P. Yu, A novel retrieval-augmented generation architecture for clinical information extraction using large language models.
- Preparing and submitting the manuscript to American Journal of Clinical Nutrition (impact factor: 5.02, Q1 journal) Z. Zhang, F. Walton, K. Lambert, C. Deng, M. Sheldon-Steemm, M. Elks, P. Yu, an ontology-based approach for understanding nutrition care needs of older people in residential aged care.
Key contact for further information
Professor Ping Yu - ping@uow.edu.au