JMIR Publications

ABSTRACT

Background:

Length measurement in young children below 18 months is important for monitoring growth and development. Accurate length measurement requires proper equipment, standardised methods, and trained personnel. Additionally, length measurement requires young children’s cooperation, making it a particular challenge during infancy and toddlerhood.

Objective:

We developed a Length Artificial Intelligence (LAI) algorithm to aid users in determining recumbent length conveniently from smartphone images and explored its performance and suitability for personal and clinical use.

Methods:

This pilot study in healthy children (0–18 months) was performed at KK Women’s and Children’s Hospital, Singapore from November 2021 to March 2022. Smartphone images were taken by parents and investigators. Standardised length-board measurements were taken by trained investigators. Performance was evaluated by comparing the tool’s image-based length estimations with length-board measurements (bias [mean error, the mean difference between measured and predicted length]; absolute error [AE, magnitude of error]). Prediction performance was evaluated on an individual-image basis and subject-averaged basis. User experience was collected via questionnaires.

Results:

A total of 215 subjects (median age 4 months) were included. The tool produced a length estimation value for 2211 (99%) of 2224 photos analysed. The mean AE was 2.47 cm for individual image predictions and 1.77 cm for subject-averaged predictions. Investigators and parents reported no difficulties in capturing the required photos for most subjects (85%, 182/215 subjects and 72%, 144/200 subjects, respectively).

Conclusions:

LAI is a practical and novel way of estimating children’s length from smartphone images without the need for specialised equipment or trained personnel. LAI’s current performance and ease of use suggest its potential for use by parents/caregivers with an accuracy approaching that typically achieved in paediatric outpatient clinics. The results show that the algorithm is acceptable for use in a personal setting, and this serves as a proof of concept for use in clinical settings. Clinical Trial: The study was registered at ClinicalTrials.gov (https://clinicaltrials.gov/ct2/show/NCT05079776)