JMIR Publications

ABSTRACT

Background:

Measurement of sodium intake in hospitalized patients is critical for their care. In this study, artificial intelligence (AI)-based imaging was performed to determine sodium intake in these patients.

Objective:

The applicability of a diet management system was evaluated using AI-based imaging to assess the sodium content of diets prescribed for hospitalized patients.

Methods:

Based on the information on the already investigated nutrients and quantity of food, consumed sodium was analyzed through photographs obtained before and after a meal. You only look once (YOLO v4)-based models and convolutional neural networks, including ResNet-101, were used to classify food and dish areas as well as food quantity, respectively. The 24-h urine sodium (UNa) value was measured as a reference for evaluating the sodium intake.

Results:

Among the 54 people enrolled, 25 participants with full data were analyzed. The results revealed that the median sodium intake calculated by the AI algorithm (AI-Na) was 2022.7 mg per day/person (adjusted by administered fluids). Although the 24-h UNa revealed a significant relationship with AI-Na along with the estimated glomerular filtration rate, the AI-Na calculations and 24-h UNa measurements differed considerably. Finally, a formula was derived using regression with an interaction term considering patients’ characteristics, such as sex, age, renal function, the use of diuretics, and administered fluids; thus, AI-Na has clinical significance in the calculation of salt intake in hospitalized patients using images without measuring 24-h UNa. Furthermore, we estimated that AI-Na corresponds to the 24-h UNa, dependent on a factor of 2.355 in the diuretics group and 0.353 in the non-diuretics group, indicating that the use of diuretics affects sodium excretion.

Conclusions:

This study highlights the potential of AI-based imaging for determining sodium intake in hospitalized patients.