Technology

Platforms. One signal path.

Everything below starts as a current at an electrode and ends as a decision. These are the places where UCARETRON does the hard part — and where OmniGram, our knowledge engine, helps design the hardware behind them.

Every platform is described in the same seven slots — what it does, why it matters, how it works, its parameters, its applications, its evidence — and each has an instrument you can turn a knob on. Those instruments compute textbook models live; none of them is measured data. How to read a claim →

Platform 01

Electrochemical Impedance Spectroscopy

Ultra-low power, ultra-compact impedance measurement. The world’s smallest, widest-band (0.1–200 kHz) Cortex‑M4 intelligent impedance chip.

A single-frequency reading tells you the electrode changed. A spectrum tells you which part of it changed — solution, interface, or diffusion — because each lives in a different decade of frequency. That separation is the difference between a sensor that drifts and a measurement you can trust.

How it works

Impedance explorer

Randles equivalent circuit · 0.1 Hz – 200 kHz

Curves are computed from a textbook equivalent-circuit model, not measured data. When analyte binds to the immobilised antibody, the electrode surface is partly blocked and R_ct rises — the widening semicircle is the signal. Drag the plot to inspect a frequency.

Frequency: —
|Z|: —
Phase: —
f_c (peak): —
ΔR_ct (signal): —

R_s — solution resistance40 Ω

Shifts the whole arc along the real axis.

R_ct — charge transfer900 Ω

Sets the width of the semicircle — the binding signal.

C_dl — double layer22 µF

Moves the peak frequency f_c = 1 / (2π·R_ct·C_dl).

σ — Warburg coefficient120 Ω·s^-½

Controls the 45° diffusion tail at low frequency.

Frequency-domain analysis

Fourier transform converts complex time-domain electrical signals into analysable frequency-domain data — like a prism separating white light into a spectrum.
Picomolar sensitivity. High-precision biosensor applications detect minute changes in electrical resistance when target proteins bind to immobilised antibodies, enabling early diagnosis before symptoms appear.
Non-linear read-out. Machine learning reads the whole spectrum — raw magnitude and phase shift together — rather than a hand-fitted circuit, which is what removes the expert from the loop.

Fourier transform converting a time-domain signal into a frequency spectrum — Time domain → frequency domain Illustrative

From electrode to inference

The same spectrum that reveals a binding event also carries the electrode’s own state. Separating the two is where the model earns its keep — and why the analysis runs on the chip rather than in a lab.

EIS biosensor detecting protein binding at the electrode surface — EIS biosensor — antibody-functionalised electrode Literature

Technical parameters

Measurement bandwidth: 0.1 Hz – 200 kHz
Measured quantity: complex Z — |Z| and phase
On-chip processor: Arm Cortex-M4
Target sensitivity class: picomolar (protein binding)
Classification accuracy: > 95 %

Applications

Label-free immunoassay — protein biomarkers
Hematocrit correction — blood glucose strips
Electrode health — drift and fouling detection
Cell and tissue impedance
Wearable contact quality

Evidence

Published spec Bandwidth, processor and measured quantity are as published for the impedance chip.
Literature Charge-transfer resistance rising on antigen capture is standard electrochemistry; UCARETRON’s own aptamer-biosensor work is published in Anal. Chem. 2026 (doi:10.1021/acs.analchem.5c07608).
Illustrative The impedance explorer above evaluates a Randles model live. It is a teaching instrument, not a data set, and no figure it produces is a specification.

Platform 02

Bio-Sensor

Electrochemical immunoassay, photothermal PCR, and cold-chain-free LAMP — next-generation molecular diagnostics designed to work outside the laboratory.

A molecular result that takes a day and a courier is a different product from one that takes twenty minutes at the bedside — even when the chemistry is identical. Getting there means attacking the two things that keep the laboratory big: the heater and the incubation.

How it works

Assay explorer

Lumped thermal model · Langmuir binding

Both tabs are computed illustrations — textbook models evaluated live so you can see which way a parameter pushes the physics. Neither is measured data and neither describes a specific UCARETRON product.

Heating ramp: —
Cooling ramp: —
Cycle time: —
40 cycles: —
T∞ reachable: —

Optical power80 %

Raises the temperature the illumination can asymptotically reach.

Nanoparticle loading70 %

The absorber sits in the liquid, so loading sets how much light becomes heat.

Thermal mass70 %

Sets both time constants. Smaller chamber, faster cycling — and less sample.

Photothermal PCR

Plasmon resonance heating. Light hitting metal nanoparticles excites surface electrons, converting energy to localised heat instantly — precise thermal cycling without bulky hardware.
Under 250 g, under 25 minutes. From sample to smartphone, wherever the sample is taken.
Multiplex. Viruses and bacteria on one platform, with laboratory-grade accuracy at the point of care.

Electrochemical immunoassay

Smartphone-integrated. Antigen–antibody reactions become electrical signals. Apply a sample to a disposable strip; read the result immediately.
Thousands of times more sensitive than standard ELISA in the sandwich format — detecting virus even during the incubation period.
CARETRIX condenses laboratory equipment into a desktop device, bringing quantitative analysis to primary clinics without large-scale facilities.

Technical parameters

Sample to answer: < 25 min
Instrument mass: < 250 g
Amplification: photothermal PCR · cold-chain-free LAMP
Immunoassay format: sandwich, electrochemical read-out
Multiplexing: viral and bacterial targets, one cartridge
CARETRIX analyser: IVD Class 1 · UCT-IMRCT-01

Applications

Respiratory panels — point of care
High-risk infectious agents — CARETRIX DYC
Field and pharmacy testing — no cold chain
Primary clinics — quantitative immunoassay
Veterinary and food safety

Evidence

Public record CARETRIX immunoassay analyser, IVD manufacturing notification Class 1, 체외 제신 25-1214 호; CARETRIX DYC kit, IVD manufacturing approval Class 3, 체외 제허 25-686 호. Manufactured under ISO 13485:2016 and KGMP — numbers and validity on the Evidence section.
Literature Plasmonic photothermal heating and isothermal LAMP amplification are established in the primary literature; this platform applies them, it does not claim to have discovered them.
Published spec Time-to-answer and instrument mass are published platform specifications.
Illustrative The assay explorer computes a lumped thermal model and a Langmuir binding curve. Its detection limits and cycle times describe the model, not a product.

Platform 03

AI Analysis

Precise biometric data processing — from fingerprint restoration to multimodal biosignal integration for security and healthcare.

Most biometric systems discard whatever is not clean. The evidence that matters is usually the part that was thrown away: a partial print, a smudged one, two prints on top of each other. Recovering structure from those is a different problem from matching a good scan, and it is the one worth solving.

How it works

Ridge lab

Procedural ridge fields · two problems, drawn

Both tabs are symbolic illustrations. The ridge fields are generated procedurally — no real print is shown — and the “recovered” panel is simply the field the generator started from. This shows what the task is, not how TruePrint does it, and makes no claim about accuracy.

Degradation: —
Usable ridge area: —

Degradationnoise · contrast · dropout

Sensor noise, contrast loss and missing patches — all at once, the way a latent print actually arrives.

Push it far enough and the left panel stops looking like a print at all. The right panel is what a restoration step has to put back.

TruePrint

Ridge restoration. AI reconstructs clear ridges from noisy, incomplete latent prints — recovering evidence traditional forensic methods cannot process.
Layer separation. A world-first technique that unmixes overlapping fingerprints into distinct individual layers.
Multimodal. Fingerprint, fundus imaging and EEG converge on one platform for next-generation security and diagnostics.

AI ridge restoration converting a degraded latent fingerprint into a clear ridge pattern — Latent print → restored ridges Product output

Multimodal integration

TruePrint Lab has been shown at major technology exhibitions, demonstrating real-time AI fingerprint extraction and analysis to industry partners worldwide. Fundus and EEG are the declared next modalities on the same platform.

Multimodal biosignal platform combining fingerprint, fundus and EEG data — Fingerprint · fundus · EEG on one platform Design intent

Technical parameters

Shipping product: TruePrint Lab v1.0
Software quality grade: GS certification, Grade 1 (TTA)
Procurement status: Innovative Product designation (KNPA)
Primary task: latent ridge restoration
Second task: overlapped-print layer separation
Declared modalities: fingerprint · fundus · EEG

Applications

Forensic latent-print recovery
Overlapped-impression casework
Public-procurement identity systems
Fundus-based screening
EEG biosignal fusion

Evidence

Public record TruePrint Lab v1.0 — GS quality certification Grade 1, No. 23-0440; Innovative Product designation for public procurement, No. 2024-034.
Literature Latent-space modelling work with UCARETRON affiliation is published in Int. J. Biol. Macromol. 2024 (doi:10.1016/j.ijbiomac.2023.127884).
Design intent Fundus and EEG on the same platform is a stated direction. The shipping, certified product is the fingerprint pipeline.
Illustrative Every ridge field in the lab above is procedurally generated. No real print appears on this page and no accuracy figure is claimed from it.

Platform 04

BEAT Connected

A connected bio-electric platform for the nervous system. It listens — heart-rate variability, nerve-conduction timing, a drawn trace — lets a model infer state, and answers with precisely phased stimulation across wirelessly synchronised nodes.

Open-loop stimulation runs to a timer and the body learns the pattern. Closing the loop means two hard things at once: measuring the nervous system well enough to decide, and delivering a pattern that never quite repeats. Both are timing problems, which is why they are solved with phase rather than amplitude.

How it works

Stimulation explorer

Phase-controlled nodes · relative conduction timing

Design intent, drawn. These two views illustrate mechanisms described in the public patent record. They are computed waveforms, not recordings, and BEAT Connected is offered for personal wellness — nothing here is a clinical claim.

Composite peak: —
Phase offset: —
Inter-node delay: —
Balance: —

Phase offset90°

Deliberately different phases — so there is no single pattern to habituate to.

Inter-node delay5 ms

Both nodes share one wireless clock; the delay between them is set, not drifting.

Balancecentred

Shifts drive between the two nodes without changing their timing.

Sense → infer → stimulate

Closed loop from heart to head. A wearable ECG sensor streams heart-rate variability; a model estimates the likelihood of an anxiety episode and, only then, shapes a cranial electrical-stimulation pattern — stimulation as a response to a measurement, not to a timer.
Stereo, phase-controlled nodes. Two or more stimulation nodes are linked and synchronised wirelessly; the controller drives them with deliberately different phases, with adjustable delay and balance — so the body is never offered one repeating pattern to habituate to.
Nerve function as a timing measurement. Simultaneous stimuli on two neural pathways are phase-shifted until the brain perceives them as one; that shift is a relative conduction time. Skin-impedance dependence is removed, so the reading reflects the nerve, not the electrode contact.
Psychological state from a drawn trace. A companion app shows an image, lets it fade, and asks the user to redraw it; reaction time, overlap and shape are analysed by AI, which then proposes coping strategies.

BEAT mechanisms interactive

Sweep frequency16 kHz

Low in the sweep the current cannot cross a cell membrane, so it reads the extracellular space. Past the β-dispersion it does — the same electrode pair, a different compartment.

Frequency: —
|Z|: —
Phase: —
Reading: —
Membrane crossing: —

Bio-Impedance Mapping is a concept, not a shipped feature. The other two are mechanisms from the public patent record, drawn as designed — computed curves, not recordings.

Technical parameters

Control loop: HRV inference → conditional stimulation
Node topology: multi-node, one wireless clock
Stimulation parameter: phase offset, delay and balance
Nerve read-out: relative conduction time between pathways
Protected mechanisms: 4 filings · 3 registered

Applications

Personal wellness — stress and sleep support
Nerve-function self-check
Drawn-trace state tracking
Research instrumentation

Evidence

Public record HRV-driven cranial stimulation KR 10-2965810; phase-controlled multi-node stimulator control KR 10-2459338; nerve-function diagnostics by relative conduction timing KR 10-2906148; drawing-based psychological analysis KR 10-2022-0069400 · PCT (pending). See the Evidence page.
Literature Low-intensity electric-field stimulation work with UCARETRON affiliation is published in Tissue Eng Regen Med. 2026 (doi:10.1007/s13770-025-00778-x).
Design intent The four mechanisms are described as designed, not as clinically validated. No efficacy or safety claim is made from them.
Illustrative The waveforms and the coincidence curve in the explorer are computed, not recorded.

Work with us

Which platform fits your problem?

Send us the measurement you cannot make today.

Product inquiry Devices, kits, specifications, availability. Technical collaboration Joint research, assay development, measurement problems. Investment & partnership Corporate development, licensing, distribution.