neos logo
Healthcare Medical Scribes · Academic Research

neos

neos is a non-invasive Class IIa medical device from Swiss company machineMD. A head-mounted headset with 200 Hz infrared eye tracking, it measures pupils, ocular alignment, eye movements and objective visual field, producing standardised examination reports.

Active Contact Sales No public API Verified by Guidaio
Overview

What is neos?

neos is a non-invasive medical device developed by machineMD AG, a Swiss company founded at the University Hospital of Bern. machineMD presents it as the world's first neurophthalmoscope: a head-mounted headset, fitted with refractive correction lenses, that uses infrared eye tracking at 200 Hz to measure how a patient's eyes and pupils respond to standardised visual stimuli.

The device presents those stimuli to each eye individually and to both together, then records the resulting eye movements and pupillary changes. Eight examinations can be completed in under twelve minutes, and shorter sequences run from three minutes. The measurement set covers binocular pupillometry, ocular alignment, fusional amplitudes, objective visual field, gaze holding and fixation, saccades and smooth pursuit. Pupillometry yields the RAPD neoscore alongside pupillography, pupil diameter and anisocoria; alignment uses an alternate cover test over nine gaze positions at plus or minus ten degrees, reporting median deviation in prism dioptres with a Hess-style chart. Results are processed in the cloud and returned as a standardised report that clinicians download from their desk.

The clinical intent is to replace subjective, operator-dependent bedside testing with quantified, repeatable data, and to bring specialist-grade neuro-ophthalmic assessment into primary and secondary care. machineMD lists target conditions including multiple sclerosis, myasthenia gravis, Huntington's, Parkinson's, progressive supranuclear palsy, Alzheimer's, brain tumours, stroke, traumatic brain injury, Niemann-Pick type C, Gaucher disease, NAION and idiopathic intracranial hypertension. Clinical partners named on the site include Inselspital Bern, University Hospital Zurich, Universitätsklinikum Bochum and Luzerner Kantonsspital.

neos is certified as a Class IIa device under the EU Medical Device Regulation by TÜV SÜD Danmark (notified body 2443), registered with the MHRA in the UK, and registered as a Class I device in the United States. One important qualification: the artificial intelligence in the shipped product consists of proprietary measurement algorithms such as the RAPD neoscore. The headline AI classification results published on machineMD's science pages come from research studies, several of which used laboratory prototypes or the separate research-use-only PRET device, not the commercial neos.

What it does

  • Measure binocular pupil response quantitatively, including RAPD neoscore, pupillography, pupil diameter and anisocoria
  • Assess ocular alignment with an alternate cover test across nine gaze positions, reporting tropia and phoria in prism dioptres
  • Map an objective visual field without requiring subjective patient feedback
  • Record saccades, smooth pursuit, gaze holding and fixation
  • Measure fusional amplitudes to assess binocular vision
  • Generate a standardised, downloadable examination report from the cloud
  • Hand the examination to a medical assistant or technician through a single interface
Audience

When to use neos / When not to

A quick filter to help you decide if neos is the right fit.

When to use neos

  • Neurologists monitoring multiple sclerosis, Parkinson's, Huntington's or myasthenia gravis who need repeatable, quantified measurements rather than subjective bedside impressions
  • Ophthalmology and neuro-ophthalmology clinics wanting to standardise pupillary and ocular alignment testing across operators
  • Primary and secondary care practices without a neuro-ophthalmologist on site, which machineMD explicitly targets
  • Clinical research teams running investigator-initiated oculometric studies, for whom machineMD provides a dedicated contact route
  • Departments that need to delegate examinations, since neos is designed to be operated by medical assistants and technicians

When not to use neos

  • Individuals or consumers: this is regulated medical hardware sold through sales contact and distributors, with no public sign-up
  • Anyone expecting software they can install or trial themselves, as no demo, free trial or free plan exists
  • Developers seeking an API or technical integration, since no API or public documentation is published
  • Clinicians expecting an autonomous diagnosis, as neos supports clinical assessment with human review and performs no automated individual decision-making
  • Buyers who need transparent budgeting up front, because no price, range or billing model is published anywhere
Get started

How to use neos

A typical end-to-end flow, from setup to results.

  1. Identify which route applies to you: Von Hoff AG distributes neos in Switzerland, machineMD sells directly in Europe, and machine MD Inc covers the United States
  2. Contact machineMD through the form on the neos page, or approach the Swiss distributor directly
  3. For research use, use the investigator-initiated studies form to discuss an oculometric study
  4. Agree commercial terms and the data processing agreement, since no self-service purchase exists
  5. Install the head-mounted device on site and fit the appropriate refractive correction lenses
  6. Train a medical assistant or technician to run the examination from the single operator interface
  7. Seat the patient and fit the headset, then launch the required examination sequence
  8. Let the device present the standardised stimuli and record eye and pupil responses automatically
  9. Retrieve the standardised report from the cloud once processing completes
  10. Interpret the quantified results as a clinician, since neos supports assessment but does not decide
Quick read

Pros & Cons

Pros

  • Turns a subjective, operator-dependent neuro-ophthalmic examination into standardised quantitative data
  • Fast in practice: eight examinations in under twelve minutes, with three-minute short sequences
  • Delegable to medical assistants and technicians, freeing specialist time
  • Certification is verifiable and nominative: EU MDR Class IIa certificate G11DK 128750 0001 from TÜV SÜD Danmark, valid to 6 October 2029, downloadable from the site
  • Quality system certified to ISO 13485:2021 by TÜV SÜD Italy
  • Unusually clear privacy posture: named subprocessors, quantified retention, and an explicit commitment not to train on patient data
  • Backed by a published body of peer-reviewed studies and named university hospital partners

Cons

  • No price is published anywhere: no pricing page, no range, no billing model
  • No demo, free trial or free plan, so the device cannot be evaluated without a sales conversation
  • No API and no public technical documentation, limiting integration into existing clinical systems
  • Requires dedicated hardware, so it cannot be deployed as software alone
  • Regulatory footing is uneven: Class IIa in the EU but only Class I registration in the United States
  • Feature availability varies by region, and the INO neoscore is available in no region at all
  • The most striking AI results come from research studies, sometimes on laboratory prototypes or the research-only PRET device, rather than from the commercial product
Pricing

Pricing & Plans

No price is published. machineMD does not operate a pricing page, and no amount, range or billing unit appears anywhere on the site, including across the forty-one non-news pages listed in its sitemap. There is no free plan and no free trial. neos is a medical device sold through commercial contact and distribution: Von Hoff AG in Switzerland, machineMD directly in Europe, and machine MD Inc in the United States. Prospective buyers must request a quotation.

Plan 1
  • No published pricing tiers
  • commercial terms are established on request
Europe
  • direct contact with machineMD
United States
  • contact machine MD Inc
  • based in Cambridge
  • Massachusetts
Plan 5
  • Research collaborations and investigator-initiated studies via a dedicated form
Special offers — No commercial promotions, discounts or reduced-rate schemes are published · Research collaborations and investigator-initiated studies are offered through a dedicated contact form, which is the only non-standard access route the site describes
Prices and plans listed above may evolve. Always check the official pricing page before subscribing.
Trust & Privacy

Data, GDPR & hosting

A consolidated view of how neos handles your data.

GDPR overview

GDPR implementation is concrete and documented. machineMD has appointed an Article 27 EU representative, MED-RAS GmbH (Eichenallee 8H, 21521 Wohltorf, Germany, ec-rep@medras.de), and names a data protection contact reachable at privacy@machinemd.com. The imprint sets out data subject rights under Articles 15 to 21 and the right to complain to a supervisory authority, or to the Federal Data Protection and Information Commissioner in Switzerland. Controller and processor roles are separated explicitly for patient data. The vendor states it carries out no automated individual decision-making or profiling within the meaning of Article 22. Website processing relies on Article 6(1)(b) and 6(1)(f) and uses Google Analytics with IP anonymisation, Google Fonts and HubSpot. The website privacy policy was updated on 2 July 2026. Note that the site never states in so many words that it is GDPR compliant.

Who owns the data?

For patient and examination data, the customer healthcare organisation is the controller or otherwise responsible party, and machineMD acts strictly as a processor, working on documented instructions under the applicable agreement and data processing agreement. machineMD may act as controller only for customer account, user, support, security, billing and contract-administration data. The customer decides what patient data is entered, manages user access, and instructs machineMD on retention, export and deletion. machineMD states it does not sell patient data, does not use it for advertising, and does not determine medical purposes or treatment decisions. On termination, data is deleted or returned under the agreement, subject to mandatory retention periods and legal holds.

Reuse rights

machineMD limits its own use of the data to operating neos and its cloud services: processing examinations, generating clinical reports, administering accounts and support, maintaining security and auditability, investigating defects, and meeting medical-device regulatory duties such as complaint handling and post-market surveillance. The vendor commits in writing that it does not use customer patient data for product development, machine learning or other secondary purposes unless the customer separately instructs it to in writing and the law and contract permit it. Because the customer remains the controller, it is the customer, not machineMD, who determines further reuse of the clinical data it generates. neos is explicitly not designed to identify individuals from gaze data or eye videos, and creates no biometric templates.

Data retention & training

Retention summary
machineMD keeps patient and examination data on the customer's behalf, following the customer's documented instructions and applicable law. Where neos is used for ordinary clinical records, retention generally matches the legal record-keeping period: at least ten years for EU deployments and at least twenty years for Swiss deployments, unless law or contract requires longer. Audit logs covering authentication, access and changes to patient records may be kept for as long as the applicable patient-record period. De-identified operational and security logs may be kept for up to three years, after which data is deleted or de-identified. Customer account and support records are kept for the customer relationship and then for statutory periods. On termination, data is deleted or returned under the agreement, subject to required retention, backup cycles and legal holds. Website server logs are deleted within seven days.
Trains on customer data
No
Subprocessors disclosed
Yes
DPA available
Yes
GDPR contact

Hosting summary

machineMD names its subprocessors, which is uncommon. Microsoft Azure hosts and stores patient and examination data, customer data and related neos cloud data. Datadog handles de-identified application, system and security logging and monitoring, and machineMD states explicitly that Datadog stores no patient or examination data. For EU, UK and Swiss customer deployments, patient and examination data are generally stored in the EU, as are the related de-identified operational and monitoring logs. No individual hosting country is named, only the EU region. Authorised machineMD personnel in Switzerland may access data where necessary for support, security and regulatory purposes, which the vendor says is safeguarded by the EU and UK adequacy decisions for Switzerland. Stated safeguards include role-based access control, least-privilege access, encryption in transit, logging and monitoring, restricted internal access, and vendor and security review processes. machineMD requires its subprocessors to protect data under written agreements.

Hosting regions
EU
Watch-outs

Things to keep in mind

Risks and trade-offs to weigh before adopting neos.

  • Do not read the published AI accuracy figures as product performance: several of those studies used laboratory prototypes or the research-use-only PRET device rather than the commercial neos
  • The regulatory footing differs by market. Class IIa certification under EU MDR is not equivalent to the Class I registration held in the United States, and buyers should confirm what applies to them
  • Feature availability is not uniform: machineMD states that the intended use and availability of the RAPD neoscore vary by region, and that the INO neoscore is available in no region
  • Quantified output can invite overconfidence. The vendor is explicit that neos supports assessment with human review and does not decide; treating a number as a diagnosis would misuse the device
  • Patient data here is health data and, through eye-region video, unusually sensitive. The customer remains the controller and carries the duty to inform patients and establish a lawful basis
  • Absence of published pricing makes budget comparison impossible before entering a sales process, and total cost of ownership for hardware, training and support is not disclosed
  • The device requires supervision in practice: an independent usability study found it is not self-explanatory, and roughly half of participants had transient goggle marks after the examination
Setup

Setup & Integrations

Technical difficulty

Setup is a procurement and training exercise rather than a technical one. There is no self-service installation: acquisition runs through machineMD or a distributor, and dedicated hardware is delivered with refractive correction lenses. Once in place, the device is designed for medical assistants and technicians to operate from a single interface, and reports are read from the cloud, so no integration work is required. Equally, no integration is possible, since machineMD publishes no API. The main effort is operator training: an independent usability study found the device is not self-explanatory and needs supervised operation by a trained technician.

Deployment

Web app
Company

Behind neos

Company name
machineMD AG
Founded
23/12/2019
Country of origin
🇨🇭 Switzerland
Headquarters
Weyermannsstrasse 36, 3008 Bern, Switzerland
US office
machine MD Inc, 1 Broadway, Cambridge, MA 02142, United States
UBO
INFORMATION_NOT_FOUND
UBO country
INFORMATION_NOT_FOUND
Domain registrar country
🇩🇩 Germany

Fundraising

Pre-Series A round announced on 18 March 2025, with Topcon Healthcare, Inc. joining as lead investor in a strategic investment; machineMD published no amount for the round
Registered share capital increased in successive steps at the Swiss commercial register: CHF 100,000 at incorporation in 2019, CHF 150,000 in January 2021, CHF 155,573 in June 2022 and CHF 379,336 in November 2022

Social

Official links

Resources

All the official URLs gathered for verification and reference.

FAQ

Frequently asked questions

What exactly is neos?
neos is a non-invasive medical device made by machineMD AG: a head-mounted headset using infrared eye tracking at 200 Hz to assess eye movements, pupils, ocular alignment and objective visual field, and to return the results as a standardised report.
How long does an examination take?
Eight examinations can be performed in under twelve minutes. Shorter examination sequences run from around three minutes.
Who operates the device?
neos is designed to be operated by medical assistants and technicians through a single user interface, rather than requiring a specialist. One independent usability study noted that it is not self-explanatory and needs supervision by a trained technician.
Is neos a certified medical device?
Yes. It is certified as a Class IIa device under the EU Medical Device Regulation by TÜV SÜD Danmark, notified body 2443, under certificate G11DK 128750 0001, valid from 7 October 2024 to 6 October 2029. machineMD's quality management system is certified to ISO 13485:2021 by TÜV SÜD Italy.
Is neos available in the United States?
machineMD states that neos is registered as a Class I medical device in the USA, under registered establishment number 3029906225, and that machine MD Inc in Cambridge, Massachusetts supports US customers. Note that this is a lower regulatory class than the EU Class IIa certification.
Where is patient data hosted?
Patient and examination data are stored in Microsoft Azure. For EU, UK and Swiss deployments, that data is generally stored in the EU. Datadog is used only for de-identified logging and monitoring and holds no patient or examination data.
Is patient data used to train AI models?
No. machineMD states it does not use customer patient data for product development, machine learning or other secondary purposes unless the customer separately instructs it in writing and the use is permitted by law and contract.
How much does neos cost?
machineMD publishes no price. There is no pricing page, no free plan and no free trial. Buyers go through the contact form or, in Switzerland, through the distributor Von Hoff AG.
Does neos make a diagnosis on its own?
No. machineMD states that neos supports clinical assessment and reporting but is not intended to make decisions about individuals without human review, and that it performs no automated individual decision-making within the meaning of Article 22 GDPR.
How long is data kept?
Retention follows the customer's documented instructions and applicable law. For ordinary clinical records this is generally at least ten years for EU deployments and at least twenty years for Swiss deployments. De-identified operational and security logs may be kept for up to three years.
Conclusion

Should you pick neos?

neos is a narrow, serious product rather than a general-purpose AI tool, and it should be judged on that basis. machineMD has built a head-mounted device that converts a traditionally subjective neuro-ophthalmic examination into quantified, repeatable measurements, and it backs that claim with the kind of evidence that is rare in this directory: a nominative EU MDR Class IIa certificate that can be downloaded and checked, an ISO 13485 quality system, named clinical partners at four university hospitals, and a substantial published research corpus.

Its data governance is also unusually explicit. Subprocessors are named rather than alluded to, retention periods are quantified, an Article 27 EU representative is designated, and the vendor commits in writing not to train models on patient data. Buyers in regulated healthcare settings will find more to work with here than on most vendor sites.

Two reservations matter. First, commercial opacity is total: no price, no range, no trial, no demo, and no API. Every path runs through a sales conversation, which is normal for medical hardware but leaves prospective buyers unable to size the investment. Second, the artificial intelligence story needs care. The shipped device runs proprietary measurement algorithms, and those are real. The eye-catching results, such as Parkinson's classification at an AUC of 0.96, come from research studies, several conducted on laboratory prototypes or on PRET, a separate research-use-only device. They describe where the technology may be heading, not what a clinic buys today.

For neurology and ophthalmology services seeking standardisation, and for research groups building oculometric datasets, neos is credible and worth a conversation. For anyone expecting a deployable AI diagnostic, it is not that, and machineMD does not claim it is.