Machine Learning and 3D Image-Based Modeling for Real-Time Body Weight and Body Composition Estimation During Emergency Medical Care. Study 1

Machine Learning and 3D Image-Based Modeling for Real-Time Body Weight and Body Composition Estimation During Emergency Medical Care. Study 1 - Establish a Model Using a Single 3D Camera Image of a Supine Patient to Accurately Estimate TBW, IBW And LBW.

Status

Withdrawn

Phases

Unknown

Study type

Observational

Source

ClinicalTrials.gov

Registry ID

NCT06646120

Enrollment

Registered

2024-10-17

Start date

2025-07-01

Completion date

2026-06-30

Last updated

2025-12-19

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

Body Weight in the Overweight and Obese Class - I Population, Body Weights and Measures

Keywords

3D camera weight estimation, computer vision weight estimation

Brief summary

The goal of this observational study is to train and validate an AI-driven 3D camera system to estimate total body weight, ideal body weight and lean body weight in male and female adult volunteers of all ages. The main questions this study aims to answer are: * What degree of accuracy of weight estimation can we achieve with an AI-driven 3D camera weight estimation system? * Is this accuracy the same in adults of both sexes, all ages, and all body types (underweight, normal weight, overweight)? Participants will undergo some anthropometric measurements (height, mid-arm circumference, weight circumference, hip circumference, measured weight), a DXA scan (to measure lean body weight), and 3D imaging using a 3D camera. There will be no interventions.

Detailed description

This study is a single-centre observational study to train, internally validate, and test an AI-driven 3D camera weight estimation system. Our hypothesis is that this system, when used in the management of acutely ill patients, will be able to estimate total body weight, ideal body weight, and lean body weight more accurately than other current point-of-care system. Healthy volunteers will be used to train and test the system. During a single data collection session of approximately 30 minutes, baseline anthropometric data, a DXA scan, and 3D camera images of volunteers lying on a medical stretcher will be captured. There will be no interventions, and no follow up of participants. The collected data will be used to train an AI algorithm (based on artificial neural networks) to estimate weight using a single depth image. Once the AI system is fully evolved, the accuracy of its weight estimation performance will be evaluated in an independent test dataset.

Interventions

None listed

Study design

Observational model

OTHER

Time perspective

CROSS_SECTIONAL

Eligibility

Sex/Gender

ALL

Age

18 Years to No maximum

Healthy volunteers

Yes

Inclusion criteria

* Any willing volunteer.

Exclusion criteria

* Participants with a body weight exceeding the DXA machine capacity \>204kg (450lbs); * Pregnant participants; * Participants with medical conditions that could confound the study; * Participants with any metallic surgical implants; * Participants who have had an x-ray with contrast in the past week; * Participants who have taken calcium supplements in the 24 hours prior to the study.

Design outcomes

Primary

Measure	Time frame	Description
TBW estimation	Baseline	Accuracy of TBW estimation using 3D camera system
IBW estimation	Baseline	Accuracy of IBW estimation using 3D camera system
LBW estimation	Baseline	Accuracy of LBW estimation using 3D camera system

Secondary

Measure	Time frame	Description
Sex-related accuracy	Baseline	Difference in accuracy between males and females
Age-related accuracy	Baseline	Accuracy of weight estimation by age-group
BMI-related accuracy	Baseline	Accuracy of weight estimation by subgroup of weight status

Outcome results