Validation Results - BestHealth Body Composition vs Hospital DEXA and Professional BIA Devices

The previous article, How to Validate Body Composition Accuracy - A DEXA Comparison Methodology, covered the validation method: which metrics to use, how to choose the cohort, and how to control the procedure. This article publishes actual comparison results from a 100-person cohort for the BestHealth body composition solution, focusing on two questions: agreement with tertiary-hospital DEXA as the gold-standard reference, and consistency with marketed BIA comparison devices. It is intended for manufacturers evaluating BestHealth modules and preparing accuracy-validation materials.

How the validation was done

On one cohort of 100 people, this validation ran same-period measurements against two kinds of reference:

• Gold standard: DEXA (dual-energy X-ray absorptiometry) at two tertiary hospitals - a commonly used reference method for body composition, used to evaluate agreement with the gold-standard reference. Using two hospitals rather than one also provides a reference for “gold standard vs gold standard” agreement.
• Professional BIA comparison devices: InBody 270 / 570 - used to evaluate consistency between the BestHealth solution and marketed professional BIA device outputs.

Each subject was measured on every device within as close a time window as possible, under controlled conditions; raw impedance and each device’s output were all logged, then Pearson correlation r, coefficient of determination and bias were computed per metric. The results below come in two parts: vs the tertiary-hospital DEXA gold standard (including dual-hospital cross-validation), and vs professional BIA comparison devices.

1. Comparison with tertiary-hospital DEXA (gold standard) and cross-validation

First, the comparison against the clinical gold standard. On the core body-composition metric of muscle / lean mass, the BestHealth solution is highly consistent with both tertiary-hospital DEXA setups; at the same time, the mutual test between the two DEXA setups gives a reproducible reference ceiling for this correlation:

A / B identify two different tertiary-hospital DEXA references. Device rows show “device - DEXA” bias; mutual-test rows show “DEXA-B - DEXA-A” bias.

The muscle-mass triptych above (left to right): BestHealth vs tertiary-hospital DEXA-A (r=0.983), vs DEXA-B (r=0.978), and the mutual test between the two DEXA setups (r=0.992).

The fat-mass triptych above (same structure as muscle mass): BestHealth vs tertiary-hospital DEXA-A (r=0.957), vs DEXA-B (r=0.954), and the mutual test between the two DEXA setups (r=0.981).

The key point: two DEXA gold standards do not agree at 1.000 with each other either (muscle mass 0.992, fat mass 0.981), and the BestHealth-vs-DEXA agreement - muscle mass 0.978-0.983, fat mass 0.954-0.957 - already sits in the same magnitude as “gold standard vs gold standard.” This means the device output tracks true body-composition changes highly consistently with the gold standard, and the correlation has a reproducible reference frame. For manufacturers preparing accuracy-validation materials, this dataset can serve as a core reference.

2. Comparison with professional BIA devices

DEXA is used to assess agreement with the gold-standard reference; professional BIA devices show how close the results are to marketed professional solutions. In this validation, comparison against InBody270 / 570 produced the following correlations:

The two figures below show fat-free mass, skeletal muscle mass and fat mass against InBody270 and InBody570 respectively (n=100); body-fat percentage is in the table above.

Fat-free mass 0.995, skeletal muscle mass 0.993-0.995, fat mass 0.988-0.989, and body-fat percentage 0.978-0.979. These results indicate high consistency between the BestHealth solution and the professional BIA comparison devices used in this validation on major outputs such as fat-free mass, muscle, fat and body-fat percentage.

How to cite the results

These results can support solution evaluation, customer technical discussions and marketing-material preparation. If you build a product on BestHealth modules, the lines below can be adapted directly into your own materials for your customers and channels - keep the sample size, comparison method and metric names together so the data scope is clear and holds up to scrutiny:

• “A body composition analyzer using the BestHealth solution, validated against tertiary-hospital DEXA (n=100): muscle mass correlation 0.978-0.983, p<0.0001 - highly consistent with the clinical gold standard on muscle / lean mass.”
• “Versus third-party professional BIA comparison devices (n=100): fat-free mass 0.995, skeletal muscle mass 0.993-0.995, fat mass 0.988-0.989, body-fat % 0.978-0.979.”
• “In dual-hospital DEXA cross-validation, the device-vs-DEXA muscle agreement (0.978-0.983) sits in the same magnitude as the two DEXA setups’ mutual agreement (0.992).”

To understand the metric choices, cohort design and procedure control behind this validation, see How to Validate Body Composition Accuracy - A DEXA Comparison Methodology; for troubleshooting correlation anomalies, see How to Troubleshoot BIA Body Composition Measurement Anomalies from Impedance Data.

For the full validation dataset, per-metric reports, or a comparison plan tailored to your target population, please contact us. Related modules: BMH05108, BMH05104-2, BMH05109.