Background: Accurate quantification of diet is essential to understand the relationship between diet and health, as well as understanding the efficacy of interventions. Yet, inherent problems from commonly used dietary recall instruments include recall issues, bias, and misreporting, which are persistent in measuring nutrients and calories. Newer ways of understanding food are emerging, particularly regarding food quality. However, common food classification systems such as NOVA fail to quantify food quality aspects. This study aimed to develop a novel food classification system that quantifies food quality using a unique scoring system combined with digital photography.

Methods: A modified food classification system based on the NOVA was developed: The Human Interference Scoring System (HISS), with quantitative measures and serving sizes used to identify the quality of digitally photographed food recalls. User testing was completed by nutritional professionals classifying five food records using the HISS to test the feasibility. The feasibility of the HISS was confirmed by identifying the inter-rater reliability using an intraclass correlation (ICC); by exploring the association between nutritional composition and HISS categories; and by conducting a thematic analysis from questionnaire responses.

Results: The HISS showed a high degree of reliability between the unprocessed (.778, p < .001) and ultra-processed (.827, p = .000) categories. The measure provided strong concordance with the usual nutrient-based measures. The ultra-processed HISS category had a moderate to strong positive association with total energy rrm (51) = .55, p < .001, total sugar rrm (51) = .79, p < .001, and carbohydrates rrm (51) = .74, p < .001, and negatively correlated with dietary fibre rrm (51), = -.39, p = .004, these were all statistically significant. The unprocessed HISS category had a weak to moderate negative correlation with total energy rrm (51) = -.14, p = .309, and total sugar rrm (51) = -.21, p = .135, and these were not statistically significant. For the unprocessed HISS category and carbohydrates, there was a moderate negative correlation rrm (51) = -.56, p < .001 and a positive correlation with dietary fibre rrm (51) = .44, p < .001, both statistically significant.

Conclusion: The HISS is a reliable and feasible instrument to use in reporting digital photography food recalls and can be used to quantify food quality by trained nutrition professionals. Energy intake, total sugar, and carbohydrates were significant predictors of foods in the ultra-processed category and quantifying food quality with the HISS was a reasonable proxy for nutrient measures. The HISS has the potential to develop nutritional epidemiology and assess the quality of dietary patterns. However, further research and development is needed to make it scalable.