The hidden factors that make body composition scans more reliable

Bioelectrical impedance analysis (BIA) is one of the go-to tools for estimating body composition — particularly when you're after total body water (TBW) or extracellular water (ECW). It's fast, portable and non-invasive. But is it accurate enough for research or clinical diagnostics in ageing populations?

One study tackled this head-on, asking a critical question: can we make body scans more reliable for elderly patients by tweaking the model behind the measurement? The answer — in short — is yes. And the solution is surprisingly pragmatic: use better variables.

The standard approach falls short

Older BIA models used simple variables like height²/resistance, sometimes weight, and a few demographic factors. But when these models were applied to elderly people, the results were all over the place:

• Some models underestimated TBW by up to 5 litres

• Others overestimated by nearly 7 litres

• Even the best of the bunch had a ~3.8% bias in healthy older adults
  

That level of inaccuracy might slide in fitness settings — but not when you're aiming for clinically actionable insight.

What actually improves accuracy and give more reliable body scans?

The researchers developed new BIA models by adding:

• Geometrical body-shape variables: wrist, mid-arm, waist and hip circumference

• Plasma osmolarity: to reflect ion concentration and fluid shifts
  

Adding these to the equation improved prediction precision dramatically — reducing the standard deviation of error from 1.8L to 0.8L.

The best-performing model explained 99.2% of the variance in TBW. That’s elite-level precision for a field tool.

Why this matters for PhantomOmics

We already know that no single signal gives the full picture — it’s the combination that matters. This study reinforces that idea. Resistance alone doesn’t cut it. But if you layer in shape data and biochemical context, you unlock more accurate models — especially in populations where body composition is shifting, like the elderly.

At PhantomOmics, our platform is built to integrate multiple, complementary signals — ECG, bioimpedance, thermography, and more. This kind of research backs our approach: measure smarter, not just more.

BIA isn’t deadweight. It just needs the right brain behind it. Read the full paper here.