A new study is exploring the potential of technology to monitor patients’ food intake.
Researchers have developed an image-based system that could help reduce malnutrition in hospitals. It is powered by artificial intelligence (AI) and machine learning.
Kabale Oke, a PhD student, is leading the project within the NIHR Southampton Biomedical Research Centre (BRC). Her work forms part of the NIHR Southampton BRC’s Data, Health and Society theme.
Kabale is guided by an interdisciplinary supervisory team at the University of Southampton. She is also working with external collaborators at the University of Reading.
Patient nutrition
Malnutrition is common in hospital patients. This is when the body does not get all the nutrients it needs to function properly.
A survey by the British Association for Parenteral and Enteral Nutrition revealed the scale of the problem in 2022. It found that nearly half of all adults screened across health and care settings in the UK were at risk of disease-related malnutrition. Patients with cancer and gastrointestinal conditions were the most likely to be affected.
The main cause of malnutrition in hospital patients is eating too little.
Studies show that poor food intake is linked to longer hospital stays, cognitive decline and a higher rate of complications. These factors contribute to increased economic costs.
AI analysis
Traditional methods for monitoring what patients eat can be time-consuming for staff. These include keeping a written food diary. In hospital settings, staff often estimate food intake based on what patients leave on their plates.
“Malnutrition in hospital patients is a significant and widespread issue,” Kabale said.
“Image-based methods offer an exciting opportunity to transform how we monitor food intake, providing accurate data with minimal user input.
“Using this technology, we hope to identify patients at risk of malnutrition earlier. This will help us ensure they receive the care and support they need without delay.”
The AI-powered system which will be trialled in Southampton is designed to be user-friendly. It consists of a camera, interface screen and a tray holder with a weight sensor and QR code.
The device will photograph patients’ plates before and after mealtimes. Using machine learning, it will generate rapid and reliable data about what patients eat.
Basil Bennett Batov is working to update and finetune the Machine Learning Model. He is also making sure that the system is maintainable and easy to use.
“The system will allow measurement of food items in real-world conditions”, Basil said. “It will reduce errors that can lead to inaccurate food intake estimates.”
Involvement of staff
Implementing new systems in hospitals requires a collaborative and interdisciplinary approach.
The research team have worked closely with Amy Popman, Dietitian for Managed Services at SERCO.
Dr Caroline Childs, Associate Professor in Nutritional Sciences and Kabale’s PhD supervisor, said:
“Successful implementation will require close collaboration with a range of stakeholders. This includes clinical staff, ward hosts and the SERCO Patient Catering Team at UHS.
“It’s great that Kabale has been able to build strong and two-way interactions with these teams. This will enrich the value and later application of her research.”
Next steps
Kabale is currently undertaking Patient and Public Involvement and Engagement (PPIE) activities. This will ensure the patient voice is embedded into her research activities.
Insights from Professor Yuanyuan Yin at the Winchester School of Art are guiding the design and user-acceptability of the device. She is part of Kabale’s supervisory team.
Future research will also explore use of the device in university catering environments. This will involve working with external collaborators at the University of Reading. They are participating in the Menus of Change initiative to assess food choices made by young people.
Comments