Digital Nutritional Assessment Tool
Lead Research Organisation:
University of Stirling
Department Name: Sport
Abstract
Malnutrition (or insufficient nutrient intake) is a significant issue for older people in the UK, and is strongly associated with poorer quality of life, functional decline and deteriorating health. Over 35% of people in care homes are affected by malnutrition with risk increasing the longer someone lives in care with it taking as little as 2 days to become malnourished. This project will develop a tool to identify malnutrition and its risk through taking photographs to identify nutritional intake. Previous studies have highlighted that malnutrition is not recorded accurately in older people in health and social care with extreme inaccuracies in diet diaries and problems including lack of staff time, difficulty remembering foods eaten, or needing a dietician to interpret the results. Linking food photos to existing food information means that it is possible to store detailed nutrient intake. We will use computer learning to link photographs of food to important measures such as food preference, physical symptoms, and malnutrition risk. The food classification tool will be delivered through a mobile phone app which will also consider portion size and eating frequency. Through this we will be able to link food intake to preferences and/or symptoms and identify foods hazardous or not preferred to decrease malnutrition risk among older people. We will develop the prototype tool and test how acceptable and usable it is in people living in care homes and their care staff, including people living with dementia.
| Description | Phase 1 Co-Development of the Tool The first step of the development of the tool was to create an advisory group (AG) with key stakeholders to develop ideas and identify features that would be key for a food recognition tool in care home setting. The study and advisory group were advertised through email to the Stirling 1000 Elders email list looking to recruit two older adults. A snowball technique was used to recruit key stakeholders working within nutrition and care settings. Through this the advisory group with 6 key stakeholders (older adults, care home staff, app developers and a dietician) met with the research team for two hours on 2 occasions over 1 year to design, test and improve the protype. They also tested the final prototype and gave feedback through a survey to replace an in-person advisory group before testing in community dwelling older adults (when an in-person date could not be arranged). AG 1: Key features for the protoype were discussed and priority was assessed. Discussion around the need for a food recognition model was had and key features of it were noted. How it would work in the care home environment, issues of data security and consent, symptoms needed to be included, as well as a mood scale and overall usability of the app were also considered. AG2: A basic prototype was developed, and food was provided to test using the tool, taking photos of different foods during the advisory group. A walk through of the basic prototype and presentation of different food recognition models was given by GameDr. The prototype was tested by taking photos of fruits and vegetables after GameDr's presentation. Basic feedback was then discussed, and further features identified. The group then tested the app with whole meals over dinner, further feedback was then given. Other features identified to enhance the app included having an after photo to calculate food eaten, adding a food fortification element, adding in medical conditions and methods to keep the tool simple such as tick box approach and symptom recording appearing when the after photo is uploaded. Due to the nature of the project LogMeal was chosen as the food recognition model provider due to the size of their database and accuracy of food recognition. This advisory group ran for twice the duration to allow for the app company to talk us through the app and food recognition models, testing in the group setting with singular foods and testing the app with whole meals over dinner. Key Design issues 1. White section/confirm screen is high up and covering the photograph 2. If photo doesn't recognise the food how is it entered- if it isn't any of the options 3. Circles sometimes did not link to a food was just circled on the plate 4. If there is an error it is difficult to return- cancel button may be needed 5. How will we distinguish plate size-what if there is a smaller portion on a big plate Key Features 1. Think about how to select Resident- ID/room no 2. Plate size option 3. Before and after photos 4. Add in a food fortification button- butter, cream, protein powder etc 5. On user profile option to add in medical conditions 6. Having tick box of various symptoms that can be personalised to patient and structure the page following the second picture or symptoms to enter later 7. Symptoms page straight after the second photo that includes fullness, mood, swallowing AG 3- Testing and feedback: The app was then refined and tested by the advisory group in free living conditions over 1 week. They were then asked to fill out a short survey on their use of the app. It found: Key Symptoms Overall, it was advised the tool monitored a good range of symptoms specific to the target group. However, some suggestions for further symptoms were thirst, confusion and anxiety. Key Features It was reported that when the app connected well with the food recognition model it was easy to use. Most of the group said that they didn't feel any key features were missing but that when testing in care home settings they will be able to highlight site-specific features. Some suggestions included an empty plate button and a directory of foods. Other Populations The survey highlighted other groups who may find this useful. These included; community care, dementia patients, family carers and hospitals. 3. Phase 2: Testing the Tool in Community Dwelling Older Adults A total of 10 community dwelling older adults over the age of 60 years were recruited through Stirling 1000 Elders email list. They were required to test the tool for two weeks and give feedback on usability and future feature development. Nine of the ten older adults completed the two-week study. One had trouble with download and decided to not participate further in the study. Key feedback was positive with eight wanting to continue using the tool on a regular basis. Other feedback included Wi-Fi and connectivity issues and issues with initial food recognition of obscure foods. These were rectified with delayed photo upload and manual food identification entry with future recognition. Issues with connecting to and interaction with Logmeal, the food recognition system that was being used, was evident in this testing phase. New Partnerships developed: Scottish Care, NHS Highland, ENRICH Scotland, Abbotsford Care |
| Exploitation Route | Key feedback on tool design and missing features will be taken forward in the further development of the tool to use in care homes. Other areas such as Care at home and individuals with visual impairments have shown interest in using the tool. |
| Sectors | Agriculture Food and Drink Digital/Communication/Information Technologies (including Software) Healthcare |
| Description | This study co-developed a prototype that allowed users to take photos of food plates and relate it to a variety of chosen symptoms. The tool was easy to use for older adults living in the community but connection to the food recognition model slowed it down and sometimes led to the tool freezing and users unable to carry on without restarting the app. The agile methodology that was used by Game Dr allowed for prototype development, involving rapid prototyping and testing. This involved three builds of the prototype, which was tested by project team and end users after each build. The process used allowed us to design, test, redesign and retest the prototype to ensure that the design and features suggested through the advisory group was usable and integral to the overall objective of the tool. It helped us build partnerships with key stakeholders and present our data at multiple conferences. This then brough forward other groups ( care at home, partially sighted individuals) who would like to use the tool. Our journey through the co-development of a digital food recognition tool has revealed both the considerable potential and areas for further refinement and enhancement. As we steer towards further funding to refine and test the prototype in care settings, the feedback and recommendations from this project will guide our refined approach, ensuring our strategies are pragmatic, actionable and robustly designed to ensure we leave no stone unturned in monitoring diet in care home residents. The next steps towards impact will be: 1) specifically testing the feasibility of its use in a care home setting with staff and residents and 2) exploring with care home stakeholders (managers and staff across organisations) potential routes to commercialisation, i.e., would they prefer to pay for a personalised database but access the app for free, or pay for use of the app. We would prefer the eventual product to be free to the NHS and social care but will explore other routes to commercialisation such as the above with privatised care which would help fund its development and improvement. |
| First Year Of Impact | 2022 |
| Sector | Agriculture, Food and Drink,Communities and Social Services/Policy,Digital/Communication/Information Technologies (including Software),Healthcare |
| Impact Types | Societal Policy & public services |
| Description | ESRC - SSSIAA Impact Accelerator Award |
| Amount | £4,470 (GBP) |
| Funding ID | STI-23/24-P0001 |
| Organisation | University of Stirling |
| Sector | Academic/University |
| Country | United Kingdom |
| Start | 07/2023 |
| End | 02/2024 |
| Description | AgeScotland |
| Organisation | Age UK |
| Department | Age Scotland |
| Country | United Kingdom |
| Sector | Charity/Non Profit |
| PI Contribution | Dissemination to members |
| Collaborator Contribution | Collaboration to recruit participants from difficult to reach groups |
| Impact | Publication from CSO project KAPS19 detailed in publications section. CSO report online |
| Start Year | 2020 |
| Description | NHS Highland |
| Organisation | NHS Highland |
| Country | United Kingdom |
| Sector | Public |
| PI Contribution | research in care homes in remote area in our future plans |
| Collaborator Contribution | assistance recruiting sites and inputting to advisory groups for further funding |
| Impact | further funding application |
| Start Year | 2023 |
| Title | DNAtool prototype app |
| Description | App to take photos of food, analyse nutritional content and link to self-reported mood and symptoms in older adults |
| Type Of Technology | Webtool/Application |
| Year Produced | 2023 |
| Impact | Small further funding award from IAASSS. Large further funding application |
| Description | Advisory Group for IAASSS |
| Form Of Engagement Activity | A formal working group, expert panel or dialogue |
| Part Of Official Scheme? | No |
| Geographic Reach | Local |
| Primary Audience | Public/other audiences |
| Results and Impact | Advisory Group meetings to inform on activities within the IAASSS small grant project and follow-on funding application |
| Year(s) Of Engagement Activity | 2023 |
| Description | GOALD and DNAtool at Stirling Social Science Festival |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | Local |
| Primary Audience | Schools |
| Results and Impact | GOALD project and DNA tool project interactive pop-ups (Nov 2023), Forth Valley College, Alloa, Stirling. Part of Stirling Social Sciences Festival. Demonstrating digital technology for use with older people to increase physical activity, social activity and dietary monitoring. Carers, activity coordinators from care homes and Leaders from Scottish Care engaged with the event and asked questions. They then agreed to support a follow-on funding application. |
| Year(s) Of Engagement Activity | 2023 |
| URL | https://www.stir.ac.uk/news/2023/10/social-science-fest-features-theatre-displays-and--discussions-f... |
| Description | Stakeholder advisory group for UKRI-funded DNA Tool project. |
| Form Of Engagement Activity | A formal working group, expert panel or dialogue |
| Part Of Official Scheme? | No |
| Geographic Reach | Local |
| Primary Audience | Professional Practitioners |
| Results and Impact | Stakeholder advisory group of NHS staff, Social care staff, and older adults for UKRI-funded DNA Tool project to co-design and test the app being developed as part of this project to measure nutrition through photographs and AI in care home residents. |
| Year(s) Of Engagement Activity | 2023 |