Neurotech On A Spectrum

To help assess and evaluate HMIs, I will introduce a new framework in this article.

Neurotechnology is complicated. Like most deep tech, it’s hard to learn, hard to understand, hard to see how the big picture fits together, hard to validate, and hard to apply.

I think one big reason that it is so complicated is because we haven’t created frameworks and language to appropriately group together or simplify some of these big ideas (we explored this further in a previous article). So, I will be introducing a new framework in this article.

The confusion in current discussions is particularly evident when we compare different technologies. There are so many factors to consider that assessing the (already complicated) tech can require a complicated background discussion.

For instance, when I talk to veterans in the field (including academics, investors, and builders), we often first discuss which factors matter to us and what our reference points are, touching on comfort, portability, cost, signal-to-noise ratio, etc. We may avoid this by focusing our discussion on a use case with inherent factor constraints (e.g., diagnostics focus the conversation primarily on medical requirements), however, even these chats require further information and a strong background.

With novices, this becomes even more arduous, as we need to introduce the different factors which *could be* relevant or important. Throughout the discussion, every time I try to bundle the factors or reference back to them, I have to reference “the group of factors we discussed earlier”, list them out individually, or use equally obtuse phrasing.

To demonstrate the relevant factors, I often compare differences in currently available tech through the image below, which intuitively demonstrates the concepts of Quality vs Adoptability.

Often, I use diagnosing Parkinson’s Disease (PD) as a use case of reference to highlight the diverse capabilities and advantages of different devices. Particularly, I discuss the likelihood of each device becoming a go-to diagnostic method. What often surprises people is that both ends of the spectrum, from high-accuracy full-scalp medical devices to mass-market smart watches, can contribute to PD diagnostics (shout out to Rune Labs).

Even though these technologies can both be used in PD diagnostics, there are obviously quite a few differences between the technologies. This example highlights that the higher accuracy from a full medical device may not win against a mass-market device which can flag earlier diagnostics automatically without any of the resources needed for the in-hospital diagnostic. Even in diagnostics, the highest channel and SNR may not be the only (or best) solution. Sometimes, the convenience and early detection capabilities of mass-market devices can be more advantageous.

By now, I hope it starts to seem clearer that for each solution, there is a large array of important factors that we must understand, consider, and balance. We need a systemic way to think about these different factors, and evaluate these technologies.

Therefore, I propose a new framework: the HMI (Human-Machine Interface) Spectrum.

This spectrum aims to put the different factors important in neurotech on a single spectrum. This primarily focuses on a tech’s validity, ability to commercialise the solution, and impact on the end user.

In determining where a technology stands, I consider these questions:

• The usefulness of the data collected: What insights are we gaining? Can this information genuinely improve our health, productivity, or well-being?

• The methods of data collection: How do these devices gather information? Is it a seamless integration into our daily routines, or does it require a more invasive approach?

• The quality of the data: In the trade-off between convenience and precision, where does each technology stand? How do we balance the need for high-quality, actionable data with the practicalities of everyday life?

• The comparison with other data collection methods: When we place these technologies side by side, what do we learn about their strengths, weaknesses, and unique contributions to our understanding of the human condition?

• The scalability of the solution: How many people can this solution help? How mass-market appropriate is this solution? What is the total size of the audience that could use this solution?

To help illustrate this a bit better, let’s look at the HMI Spectrum applied to a common question: how clinical vs mass market appropriate is a device? One can intuitively think of some factors which may be important to making a device easily usable by a lot of people, vs a device which would be used e.g., for diagnostics in a hospital. This is a great example of a question where our Spectrum would be useful to score and rank different devices.

Expanding from this, we can look at how different current tech *may* fit into the spectrum (obviously depends on the specific device and a lot of other things). We will dive more into examples in an upcoming post, where we will walk through how we placed some of these systems according to the above questions.

This spectrum is still in its early days, but I think solves a fundamental need of giving us a shorthand to describe these important factors in tech assessment and comparison.

What important factors do you think I am missing? Where do you think the HMI Spectrum could be useful?

My name is Abby Holland. I am a Product Owner at IDUN Technologies, a consumer neurotech company building EEG-augmented headphones. I have been building in the space since 2018, and have a background in BioMechanical Engineering and Neurological Disorders. Follow me on Twitter @holland_neuro for realtime commentary on new neurotech developments.