The five stages from concept to pilot builds: which prototypes each stage needs, in what quantity, and how regulation shapes them.

By Sebastian Stanley. First published 20 November 2024. Updated October 2026 by Tom Davies-Kent, Director, Complete Fabrication.
Medical device prototyping moves through five stages: concept, design development, formative testing, verification and validation, then pilot builds and transfer to manufacture. Each stage needs a different kind of prototype, in a different quantity. A proof of concept might be one rough rig. A validation study can need tens of identical, production-like devices.
We make prototypes and small batches for medical device teams, so the notes below come from the workshop side, including where projects tend to lose time.
A study of medical device development drew on interviews with more than 80 experts in device development, commercialisation and regulation. They found that most device companies use a stage-gate process, and that regulation plays a substantive role in shaping it (Pietzsch et al., 2009). The names vary between companies, but the shape is consistent.
| Stage | What it has to prove | Typical prototype | Typical quantity | Processes that suit it |
|---|---|---|---|---|
| 1. Concept and feasibility | The core idea can work | Rough works-like rigs; simple form models | 1 to 3 | SLA, FDM |
| 2. Design development | The design meets its requirements | Separate looks-like and works-like prototypes, many iterations | A handful per iteration | SLA, Carbon DLS, CNC machining |
| 3. Formative testing | Users can operate it safely; parts fit and work together | Integrated looks-like, works-like prototypes | 5 to 20 | SLA, Carbon DLS, vacuum casting |
| 4. Verification and validation | It meets every requirement, and the intended users can use it | Production-representative devices | Tens | Vacuum casting, Carbon DLS, CNC machining |
| 5. Pilot and transfer to manufacture | It can be made repeatably at volume | Pilot builds, bridge batches while tooling is made | 10 to 500 | Vacuum casting, Carbon DLS, then injection moulding |
These quantities vary by project. A device with several user groups, for example, needs more units for usability testing than one with a single user group.
The first prototypes test whether the idea can work at all, so they are deliberately rough. A works-like rig might be a mechanism on a plate, with no housing at all.
Speed matters more than finish here. SLA 3D printing turns a CAD file into a part in 24 hours as standard, and FDM is cheaper still for large, simple forms. Keep the design open, because this is the cheapest point to change direction.
This is where most iteration happens. Teams usually split the prototype into two kinds:
Each kind can change quickly without holding up the other. When they come together, you have a looks-like, works-like prototype, as we made for Hayter's two working prototypes.
Process choice follows the job. SLA holds ±0.15 mm and captures fine detail. Carbon DLS gives functional parts with properties close to injection mouldings. CNC machining to ±0.10 mm suits metal parts, clear acrylic and anything that must be right first time.
Formative studies test the design with real users while it can still change. A review in Applied Ergonomics found that capturing user requirements is essential to safe medical devices. Yet developers had little published guidance on how to do it (Martin et al., 2008). Regulators now publish that guidance. In Great Britain the MHRA's human factors guidance treats usability engineering as iterative: design, test, and improve (MHRA, 2021).
Formative prototypes need to look and feel right, because users react to what they hold. An IPL hair removal device we made for focus groups is a typical example: 3D printed models used to assess ergonomics before the design was fixed.
Verification tests the device against its technical requirements. Validation tests whether the intended users can use it, in its intended environment. For summative usability testing, the US FDA expects at least 15 participants per distinct user group (FDA, 2016). With two or three user groups, that quickly means tens of identical devices.
Those devices have to be consistent. If one unit's button is stiffer than another's, that difference shows up in the study results. That is why validation batches usually move away from one-off prints to vacuum casting. One silicone mould makes around 20 matching castings, in polyurethanes that behave like production plastics.
The last stage proves the device can be made repeatably. In-house pilot builds or bridge batches keep testing going while production tooling is made. For many devices, the final medical devices manufacturing process is injection moulding, often by a contract manufacturer.
Bridge batches can carry a product further than people expect. Our Owlstone breath analyser casework started as one- and two-off prototypes. Over five years it grew into vacuum cast production runs of 10 to 30, which bridged the gap between trials and higher-volume manufacture.
Device regulation applies to the manufacturer and the finished device, but it reaches back into prototyping. Four frameworks come up on almost every project:
| Requirement | What it covers | What it means for prototypes |
|---|---|---|
| Design controls | Planning, inputs, outputs, reviews, verification, validation and design transfer | Record which prototype tested which requirement |
| Risk management (ISO 14971) | Identifying hazards and controlling risks across the life of the device | Prototypes are used to test risk controls |
| Usability engineering (IEC 62366-1) | Designing for safe use, with formative and summative evaluation | Usability batches must match the design under test |
| Biocompatibility (ISO 10993-1) | Biological evaluation of materials that contact the body | Choose prototype materials with the contact type and duration in mind |
In the US, the FDA's Quality Management System Regulation took effect on 2 February 2026. It incorporates ISO 13485:2016 by reference, including its design and development requirements (FDA, 2026). In Great Britain, the MHRA regulates medical devices.
A prototype is not a finished medical device, and approval stays with the device manufacturer. What a prototype supplier can do is make the evidence easier to collect. We supply Certificates of Conformity for manufacture and for materials when you flag the need at quote. For each batch we also record the materials, pattern and mould numbers, who made the parts and the check results.
Many prototypes never touch a patient. Looks-like models for design reviews can be made in any resin that gives the right appearance. Prototypes for skin-contact testing or surgical trials need materials with biocompatibility data.
We run certified biocompatible materials on our Carbon DLS printers, such as Carbon MPU 100 and SIL 30, and offer equivalent resins for vacuum cast batches. We supply the material data for your own assessment. Whether a material is suitable depends on how long it contacts the body and where, so that decision stays with your team.
Problems found late cost the most. Thomke and Fujimoto called the answer "front-loading": shifting problem-finding and problem-solving to earlier stages of development. They argued this can reduce development time and cost (Thomke and Fujimoto, 2000).
For medical devices, that means checking design for manufacturability before verification builds start. Uneven walls, undercuts that need side actions, and tight tolerances on cast parts are the usual problems. Our design for manufacture guide covers the rules. We also check every file before we make it and say if a change would make the part stronger, cleaner or cheaper.
We make appearance models, functional prototypes and trial batches for medical device teams from our workshop near Cambridge. Most medical work runs under an NDA. We choose the process for the stage. SLA suits detail, Carbon DLS repeatable functional parts, vacuum casting batches, and CNC parts that must be right first time.
Our SAMBA II work for Diagnostics for the Real World ran from the first working prototypes into production of about 20 to 30 finished units. See our medical device prototyping service for materials, finishes and certificates.
Most teams follow five stages: concept and feasibility, design development, formative testing with users, verification and validation, then pilot builds and transfer to manufacture. Each stage needs a different type and quantity of prototype.
A looks-like prototype tests appearance, size and ergonomics. A works-like prototype tests function, such as a mechanism or fluid path. Later in development the two are combined into one looks-like, works-like prototype.
It depends on the stage and the number of user groups. Formative studies often use small groups. For summative testing, the FDA expects at least 15 participants per distinct user group. A device with several user groups therefore needs tens of identical units, plus spares.
Only if they will contact a patient or user in testing. Looks-like models for design reviews do not. For skin-contact or surgical trials, choose a material with biocompatibility data that suits the contact type and duration.
Approval stays with the device manufacturer, because prototypes are not finished devices. A supplier can make the evidence easier to collect, for example with Certificates of Conformity for manufacture and materials, and batch records.
Send your CAD files, the quantity and the stage you are at, and we will recommend a process and give you a cost and a timeline. Call +44 1223 836837, email info@compfab.co.uk, or contact us.
Send your CAD files or drawings and tell us when you need the parts. Call +44 1223 836837 or email info@compfab.co.uk.
Contact us