The Next Leap in Diagnostics Depends on the Plumbing, Not the Chemistry

The pumps, valves, and algorithms that move a drop of fluid decide whether a test ever reaches a patient.

When people talk about the microfluidics revolution, they talk about the assay. The chemistry that detects a virus, a tumor marker, or a metabolite from a single drop of blood. That is the visible part.

I have spent years watching this field up close. The real bottleneck between a good lab result and an affordable clinic test is the plumbing.

Moving microlitres and nanolitres of fluid through a cartridge, precisely, on cue, at a cost that works for a disposable device, is hard. Every pump, every valve, every electrical contact is another part to manufacture, another failure mode, another fraction of a cent. Multiply that across millions of cartridges, and it decides whether a test reaches the people who need it. That is the problem we work on at Ginolis.

Three developments point to where affordable diagnostics is heading. Two in how we move the fluid. One in how we make sense of it.

Professor Kari Ullakko is a Finnish materials scientist widely credited with discovering the magnetic shape memory effect. Certain alloys, most famously Ni Mn Ga, physically change shape when exposed to a magnetic field, straining by as much as six percent.

What he and his colleagues built from that physics is one of the simplest ideas I have seen in active microfluidics. A micropump made of essentially one component, the shape memory element itself, with no membrane, no motor, no mechanical valve in the fluid path. A moving magnetic field creates a small cavity inside the alloy that travels as the field moves, pushing fluid ahead of it, similar to a peristaltic pump squeezing a tube. Rotate a small magnet outside the cartridge and the fluid moves.

The same element that drives the fluid can also block the channel to hold it in place, so one part does the job of two. No electrical connections run into the fluid, since the actuation comes from an external magnet, so the cartridge stays simple. Reported performance is on the order of 50 to 150 nanolitres per cycle, at pressures well beyond two bar, with bidirectional flow. It has also been shown not to inhibit PCR, so it works for molecular diagnostics too.

When the active fluidics live in a reusable instrument, and the disposable cartridge gets simpler and cheaper, the economics of point-of-care testing shift in the right direction.

The second approach comes from a different angle, one our own team at Ginolis explored and patented some years ago, US 10,309,926 B2, On-chip control of fluids using electrodes, by Fredrik Jansson and Magnus Molin.

The idea is to control fluid with electric fields acting on the fluid's own surface. A patterned working electrode and a reference electrode contact the sample as it travels along the channel. A hydrophobic barrier sits at the right point. With no voltage applied, it acts as a wall and the fluid stays put. Apply a potential, the surface becomes wettable through electrowetting, and the fluid advances. Switch it off, and it stops again.

An electrically actuated valve with no moving parts, addressable by software.

We have a short demonstration video. A blood sample loaded into a transparent chip, wired to a controller. A computer sequence takes over, and the blood moves down the channel, fills a chamber, and continues to the outlet, step by step, under electrode control. No pump whirring. No mechanical valve clicking.

Electrodes can be patterned at scale. The valve is geometry plus software. For high-volume, low-cost cartridges, that matters a great deal.

Two different physics, magnetic shape memory and electrowetting, aimed at the same problem. Cut the cost, moving parts, and complexity of moving a drop.

Getting the fluid to the right place at the right time is half the story. What you do with the signal that comes back is the other half.

Coupling microfluidic data with AI produces diagnostic platforms that can self-optimize and are increasingly integrated with patient records and dosage-recommendation algorithms. In oncology, machine learning applied to exosomal microRNA signatures, metabolomic profiles, and Raman and infrared spectra is turning liquid biopsy readouts into classifications a clinician can act on. Microfluidics generates rich, high-dimensional data from a single small sample. AI is the layer that makes that data usable at scale.

Smart actuation gets the right fluid to the right place. Smart data turns the resulting signal into a decision a clinician can act on. The diagnostic cartridge becomes less of a passive strip and more of a small, intelligent system.

Ullakko's micropump, the electrowetting valve, and AI-driven readouts are good engineering. Good engineering in a lab does not lower anyone's healthcare bill. An innovation matters to a patient only when it can be manufactured precisely, repeatably, at volume, and at a price that survives contact with the real world.

High-precision liquid handling, assembly automation, and quality inspection that carry a microfluidic device from an R&D bench to scaled, compliant, affordable production. The clever pump and the clever valve are only as useful as our ability to make millions of them, identically, without driving up the cost.

The technologies are arriving. Making the latest diagnostics accessible to everyone at a reasonable cost is as much a manufacturing mission as a scientific one. The companies that get this right will treat manufacturing as part of the invention.

Kauko Väinämö, CEO, Ginolis.