An IVD Is a Medical Answer, Not a Machine

An in vitro diagnostic sells a result. The reader, the cartridge, the pipette, and the app are all delivery mechanisms for one number or one call, and FDA reviews the accuracy of that call, not the elegance of the hardware. This trips up hardware teams constantly. You can build a beautiful benchtop analyzer and still have no product, because nobody defined which analyte, in which sample matrix, at which cutoff, for which clinical decision.

The practical consequence is that an IVD program is run by the assay, and the instrument is engineered around it. Reversing that order is the single most expensive mistake in this category.

Step One: Write the Intended Use Statement

Before any CAD, write one paragraph that names the analyte, the specimen type (whole blood, plasma, nasal swab, saliva, urine), the population, the clinical claim (screening, aid to diagnosis, monitoring, confirmatory), and the setting (central lab, physician office, home). Every downstream decision is derivable from that paragraph, including your regulatory class. Whether your product is even an IVD, and which class it lands in, follows the logic in FDA medical device classes.

Setting matters more than founders expect. A test intended for a physician's office or a pharmacy needs a CLIA waiver, which requires demonstrating that untrained operators get results equivalent to trained ones, plus fail-safe design so an invalid run reports as invalid rather than as a number. A CLIA waiver study runs across multiple sites with real intended operators and typically adds six to twelve months and a six-figure budget on top of the clearance work.

Assay and Instrument Are Co-Developed or Neither Works

The chemistry sets the hardware requirements, and the hardware constrains the chemistry. Some of the couplings that decide the architecture:

  • Detection method. Lateral flow with a reflectance reader, fluorescence, electrochemical, or optical absorbance each drive a completely different optical and electronics stack, and a different limit of detection.
  • Thermal control. A molecular assay needing 40 cycles between 95 and 60 degrees C with plus or minus 0.5 degree uniformity dictates the heater, the sensor placement, the consumable's wall thickness, and much of the power budget.
  • Fluidics. Passive capillary flow is cheap and unpowered but limited; active pumping and valving costs money and adds failure modes. The cartridge is usually the hardest part to industrialize.
  • Reagent stability. Lyophilized on-cartridge reagents let you ship at room temperature and dominate cartridge geometry, sealing, and moisture barrier selection.
  • Calibration and controls. Lot-to-lot variation is handled with calibrators, onboard controls, and often a barcode or RFID that carries lot-specific coefficients into the instrument.

If the instrument computes or interprets a result, the software carries its own regulatory weight. A companion app that presents or interprets output may itself be regulated, which is the territory covered in software as a medical device.

Validation Is Where the Money Goes

Analytical validation proves the test measures what it claims. That means precision studies (typically the 20 days, 2 runs, 2 replicates design), limit of blank and limit of detection, linearity across the reportable range, interference testing against hemolysis, lipemia, bilirubin and common medications, cross-reactivity against organisms or analogs, matrix comparison, and real-time plus accelerated stability for both reagents and the instrument.

Clinical validation then proves the result is clinically meaningful, usually by comparing your test against a reference method on prospectively collected specimens across several sites, reporting sensitivity and specificity with confidence intervals. Rare-condition tests are hard here simply because positive specimens are scarce; banked and contrived samples help but rarely satisfy a reviewer on their own. Budget realistically. For many IVDs, validation exceeds the entire engineering spend, a pattern visible in the ranges discussed in medical device development cost.

Keep the distinction between design verification and validation clean throughout, because reviewers do. The framing in verification vs validation maps directly onto the assay and instrument workstreams.

The Regulatory Route

Most IVDs enter through a 510(k) against a predicate with a comparable intended use, following the path described in the FDA approval process for medical devices. Genuinely novel analytes with no predicate often go through De Novo, which sets the classification for everyone who follows; the mechanics are in the FDA De Novo pathway. High-risk applications such as blood screening or companion diagnostics require a PMA. A pre-submission meeting is close to mandatory in this space, because FDA feedback on your validation protocol before you run it can save a year of repeat studies.

Manufacturing Is Two Supply Chains

You end up running an instrument supply chain and a reagent supply chain simultaneously, with different rules. Instruments are electromechanical builds with normal contract manufacturing dynamics. Reagents and cartridges require controlled environments, lot release testing on every batch, cold chain in many cases, and expiry dating that constrains how much inventory you can hold. Both sit under one quality system, and ISO 13485 needs to be in place before design transfer, not after.

The business model usually follows the razor and blade shape: instruments placed at low margin or on reagent rental agreements, with margin earned on consumables. That makes cartridge cost at volume a strategic number from day one, not a late optimization.

Funding the Validation Gap

The stretch between a working benchtop assay and a cleared product is where diagnostics companies stall, because the spend is large and there is no revenue yet. Non-dilutive money fits this phase well, and diagnostics are a strong match for the programs described in NIH grants for medical device startups.

Build the Assay and the Instrument Together

Projects House builds IVD instruments and consumables alongside assay teams, covering optics, thermal and fluidic design, cartridge design for manufacture, embedded and application software, and the verification work that feeds the submission. Send your intended use statement and current assay status through our contact form and we will scope the development and validation path.