Back

Multiplexed visual test with readout synchronized to each respective clinical threshold for rogue anti-cytokine autoantibodies

Shafique, H.; Bernier, S.; Ng, A.; Roussel, L.; Vinh, D. C.; Juncker, D.

2026-08-04 public and global health
10.64898/2026.08.02.26359144 medRxiv
Show abstract

Rapid tests with visual readout can quickly help identify and triage at-risk patients. However, multiplexed visual tests (MVTs) for targets with clinical thresholds above the limit of detection -- e.g., rogue autoantibodies (raAbs) -- are lacking because test interdependency and cross-reactivity make optimization intractable. Here, we introduce a conceptual and experimental framework to synchronize visual readout and clinical thresholds for multiple, cross-reacting targets simultaneously, and illustrate it with a 3D-printed, structurally-preprogrammed capillaric MVT for anti-interferon (IFN)- and -{omega} raAbs and anti-SARS-CoV-2 spike protein (anti-SCoV2) antibodies. Using design of experiments, we sought and identified parameters that collectively govern the background of all tests (buffer composition, ionic strength), and ones that individually govern assay signal and sensitivity (capture probe density, sample volume), thus enabling both collective background reduction and independent tuning of test line visual threshold. The instrument-free MVT is highly sensitive (pg-ng mL-1), reproducible (CV<10% in plasma), and completed in <1 h. We benchmarked the threshold-calibrated MVTs to microplate ELISA with 41 COVID-19 patient plasma samples yielding ROC-AUCs of 0.97, 1.00, and 0.98 for anti-IFN-, -IFN-{omega} and -SCoV2 tests, respectively. The proposed framework for synchronizing multiple visual readouts with respective clinical thresholds, combined with capillarics, opens the door to instrumentation-free MVTs for point-of-care use.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.