JMIR Publications

ABSTRACT

Background:

Currently over 4’000 bariatric procedures are performed annually in Switzerland. To improve the outcome, patients need to have a good knowledge regarding postoperative nutrition.To potentially provide them with knowledge between dietetic consultations, a health bot (HB) was created. The HB can answer bariatric nutrition questions in writing based on artificial intelligence (AI).

Objective:

This study aimed to evaluate the HB for its usability and perception among bariatric patients.

Methods:

Patients before or after bariatric surgery tested the HB. A mixed-method approach was used, consisting of a questionnaire and qualitative interviews before and after testing the HB. Amongst others the dimensions usability, usefulness, satisfaction, and ease of use of the HB were measured. Data was analyzed using R Studio and Excel. Interviews were transcribed according to Kuckartz and a summary inductive content analysis was performed.

Results:

A total of 12 patients (8 females, 4 males) were included. The results showed an excellent usability with means usability score of 87 (sd 12.5; range: 57.5-100) out of 100. Other dimensions of acceptability included usefulness (5.28/7, sd 2.02), satisfaction (5.75/7, sd 1.68) and learnability (6.26/7, sd 1.5). The concept of the HB and the availability of reliable nutrition information were perceived as desirable (5.5/7, sd 1.64). Weaknesses were identified in response accuracy, the HB's limited knowledge, and design.

Conclusions:

The HB's ease of use and usability were evaluated positive response accuracy, topic selection, and design should be optimized in a next step. The perceptions of nutrition professionals, the impact on patient care and nutrition knowledge of participants need to be examined in further studies.