LB034 - A DIGITAL SOLUTION TO EFFECTIVELY IMPROVE MALNUTRITION MANAGEMENT

LB034

A DIGITAL SOLUTION TO EFFECTIVELY IMPROVE MALNUTRITION MANAGEMENT

E. G. Gerrits^1,2,*, M. Ruehlin²

¹Clinic for Internal Medicine, ²Center of Nutritional Medicine and Therapy, Cantonal Hospital of Winterthur, Winterthur, Switzerland

Rationale: A substantial number of patients do not cover their daily nutritional needs by hospital food. In 2003 we introduced a plate diagram to roughly identify patients with insufficient nutritional intake. This semi-quantitative estimation and documentation by the nursing staff and the dietitians in the patient file is time-consuming and inaccurate. Therefore, digital approaches are needed to estimate the daily calorie and macronutrient count of our patients. The aim of this study was to validate an artificial intelligence-based food tracker on its accuracy and reliability in a clinical setting.

Methods: In this pilot-study, a food tracker was tested in the Cantonal Hospital of Winterthur in Switzerland, over a 3-month period, for its integration in the hospital kitchen process and for its reliability. Nutritional accuracy was measured by comparing the food tracker`s calculation of the main meals of randomly chosen 23 patients, to the caloric value and distribution of macronutrients estimated by our dietitian.

Results: From 10.12.2025 until 19.03.2026, 166’602 main meals of 15’138 hospitalized patients were scanned. The integration of the food tracker in our hospital kitchen proceeded seamlessly with only 2 minor technical incidents, which have been resolved withing 12 – 16 hours. During this period, the food tracker showed a 98.0 % identification accuracy of the plates (less than 2% of the trays were missed during the scanning process due to occlusion or missing barcodes). Nutritional accuracy without calibration was 89.2 % in macronutrient estimation and 95.7 % with calibration.

Conclusion: Digitalisation of the plate diagram of hospitalized patients, by an artificial intelligence-based food tracker, is a practically feasible, reliable and highly accurate method to estimate patient`s food intake in the hospital. It substantially allows to improve today`s malnutrition management.  

Disclosure of Interest: None declared