Table 5d.

Portable microfluidics-based methods: device performance in human blood.

Portable device vs. reference	EE-µPAD, vs. ELISA	Tidbit with HYPER platform, vs. ELISAa	Tidbit without HYPER platform, vs. ELISA
Vitamin A biomarkerf	RBPb	RBPb	RBPb
Sample type	Whole blood	Whole blood	Serum
Study population	95 adults (commercial)	12 adults	43 adults (commercial)
Concentration difference	NR	NR	NR
Correlation coefficient	NR	NR	0.75
R2 (index, unless specified)	NR	Index: 0.81 vs. ref: >0.99	0.56
Regression equation	NR	Slope = 0.99	Slope = 0.97
RMSE, index vs. ref	NR	3.75 vs. 1.3 µg/mL	4.34 µg/mL vs. NR
Operational range	∼10–70 µg/mL (graph)	∼5–20 µg/mL (graph)	2.2–20 µg/mL (0.10–0.95 µmol/L)
VAD or VAI (%), index vs. ref	<16.3 µg/mL: AUC = 0.7139 vs. 17.2%	NR	<14.7 µg/mL (≤0.70 µmol/L): NR vs. 9.3%
Precision
Sensitivity	75% at MFR cutoff, 0.831	NR	100%
Specificity	62.3% at MFR cutoff, 0.831	NR	100%
Intra-assay %CV	10.8% vs. 3.9%	20.3% deviation per test strip, recommend taking average of 3 test strips	NR
Inter-assay %CV	NR	NR	NR
Inter-observer %CV	NR	NR	NR
Bland Altman analysis comments	NR	NR	Bias at −0.05 µg/mL (-2.3 nmol/L)
Reference	(Lee et al., 2016)	(Lu and Erickson, 2017, Lu et al., 2017, Lu et al., 2018)	(Lu and Erickson, 2017, Lu et al., 2017)

Notes:MD, mean difference; MFR, multi-faceted ratio i.e., the ratio of the light transmission in the test area to that in the background control area, calculated for RBP for each sample repeat. NR, not reported; RE, retinol equivalents defined as the sum of retinol and retinyl esters, equal to 3.3 International Units (IU) of vitamin A or as 1 µg (units reported by manufacturer—however, retinol activity equivalents (RAE) are the preferred unit for reporting (Institute of medicine, 2001); RMSE, root mean squared error; VAD, vitamin A deficiency; VAI, vitamin A insufficiency.

^aReference ELISA utilized samples that were filtered using HYPER system.

^bUnits: µg/mL, mg/L, or µmol/L.