Hiss App Evaluation: AI Food Classification for Nutritional Literacy

Issue addressed: High consumption of ultra-processed foods (UPFs) is linked to poor health outcomes, yet consumers often struggle to recognise and interpret food processing. Digital tools using artificial intelligence (AI) can support nutritional literacy and UPF awareness. This study evaluated a HISS (Human Interference Scoring System)–based mobile application designed to classify foods by processing level and support dietary self-reflection. Methods: A three-day quantitative usability study was conducted in New Zealand. Thirty-one participants (13 adolescents aged 12–18 years, eight tertiary students aged 19–25 years, and nine Māori and Pacific health coaches) logged all meals, snacks and beverages using the HISS app. AI classification accuracy was assessed against expert ratings of food images. App engagement was measured using in-app metrics, and usability and perceived impact were assessed via surveys. Results: The AI system achieved 93% accuracy for HISS category classification. App engagement varied across features, with most time spent on meal logging and AI interaction screens. Adolescents and health coaches reported high usability and usefulness, while tertiary students expressed more mixed intentions regarding ongoing use. Conclusions: The app demonstrated high classification accuracy and was generally well received, particularly among users with lower baseline nutrition literacy. Findings support the feasibility of using AI-enabled image recognition to support awareness of UPF intake. So what?: Evidence based AI food classification tools such as HISS show promise for scalable UPF reduction. With further development and evaluation, such tools may support nutrition education and behaviour change in community and clinical settings.