Why Do We Need The Stack Model?

Written by Team DEEP Measures | Jan 14, 2026 12:38:58 PM

We rarely think about what makes routine health checks (like a simple blood‑pressure reading) so effortless. Clear definitions, shared standards and trusted tools make it possible. Digital measures don’t yet have that foundation. This article introduces how the Stack Model brings structure and consistency to new digital health measures, helping them become as reliable and interpretable as long‑established clinical metrics.

Imagine visiting a doctor in a new country to have your blood pressure checked. The doctor picks up a device, takes a few readings and then explains whether there’s anything to worry about - and perhaps what steps to take. It feels like a simple, everyday interaction and it is. However, for this to be such a routine event, the following needs to be true:

There needs to be an understanding of what blood pressure is so that you don’t need to debate that with the doctor. There’s been hundreds of years of history behind this measure, dating back to 1700’s. It became more standardized in the early 1900’s.
The doctor (and patients too) need to be able to make decisions based on this. Without this, the readings would have no clinical utility. Even though this is still evolving, there's quite well established clinical practice around BP so any doctor in the world should be able to help guide useful interventions.
In practice it doesn’t matter very much which particular device the doctor happens to use to measure your BP. Several different brands of devices have been cleared for this purpose and they all more or less do a fit-for-purpose job. Sure, there are different modalities and these have come a long way from the first method of measuring BP in a horse by inserting a brass pipe into the carotid artery connected to a 3-meter glass tube where blood rising and falling could be observed.

In the context of digital measures, some things are different. Often the measures we work with are novel and do not have many years of history behind them. So defining them and their interpretation needs to be established. There may be limited information about what is fit-for-purpose for measuring the new concept. We all know how rapidly digital technologies are evolving, especially with AI that may be constantly changing and it can be a challenge to manage that lifecycle in a way that still allows us to deliver consistent health data for different purposes.

So how does the stack model help address this?

The stack model offers a structured way to organize all the relevant information around a measure as well as the modalities (e.g. the tech-agnostic method) and technologies (e.g. a specific solution) to capture it.

The Measurement Definition block - what to measure and why

The Measurement Definition (MD) block addresses 'WHAT' should be measured and 'WHY', including its interpretation. The MD includes the purpose and context of use (e.g. specific population). It also includes the variables and how the measure is calculated. If we take, for example, the measurement of coughing bouts in COPD, we need to define 'what is a bout'. We can then measure many different aspects (concepts of interest) of those bouts, for example their frequency, duration, intensity etc.

Some of these concepts of interest will be more or less meaningful to a specific population and purpose of use. For example for drug development, the selected measure should be highly meaningful to patients and also relevant for the intervention being developed. These 'meaningfulness' and 'relevance' definitions can then be endorsed by regulators and the clinical community to achieve the a common understanding (as in the blood pressure example).

What the MD block does is turn these definitions into 'building blocks' that can be cataloged and repurposed for different contexts and purposes of use and also enables consistent decision making.

The Target Solution Profile block - requirements for how to measure

The Target Solution Profile (TSP) block ensures that we can define common characteristics and performance requirements for the different ways of measuring a particular measure. For example, coughing bouts could be measured based on sound or movement or both.

These measure methods (modalities) can have different form factors, for example a chest patch vs. wrist-worn device, which are two different ways to measure the same concept. They would have a different TSPs as they have different defining characteristics and may have different measurement properties.

One modality could be better for a specific use case and this is where we can choose the most relevant one, but still rely on the measure definition for consistent interpretation.

TSPs address the issue of rapidly evolving lifecycles of the different devices as TSPs are agnostic to the specific technology. There can be many different brands of chest-worn patches and there can be many different versions of those patches from the same brand.

A clinical development program can easily take a decade to execute and during this time there could be several generations of the same modality of technology. New modalities are often born within that timeframe. Wait long enough and a particular modality may even be deprecated (think of the brass pipe solution for measuring BP).

TSPs allow for consistent evidence to be generated across modalities and technologies. Similar to MD’s, TSPs can be endorsed so that we can choose the most suitable device for any given use case. This is why you don’t need to worry about the BP measurement devices in the doctor’s office, you and the doctor have confidence that it meets the required characteristics and performance requirements.

The Digital Measurement Solution block - the technology to measure

Digital Measurement Solutions are the actual devices that measure according to a TSP. They have diverse characteristics and the specific versions have short lifecycles, with software and hardware updates that can be very frequent.

For many use cases like clinical trials, it is important to choose the right technology for the trial. TSPs guide the requirements on how to measure, but what device is right for a study depends on many factors. As long as the DMS aligns with the TSP requirements, it should be fit for purpose.

So how do we ensure that a DMS meets TSP requirements?

This is where Qualification Protocols (QPs) play a role - instead of qualifying specific devices, the Stack Model offers a mechanism to qualify the qualification method instead, making it easier for people to demonstrate this.

The QPs are series of performance requirements to demonstrate that a DMS fulfils the requirements of the TSP. They might be binary, for example; “The device must have Bluetooth connectivity”. They might be about performance, for example; “The algorithm error rate must be below 10% when measured against a standardized reference dataset Y”. In this case, we would need to perform a test run and supply the results as evidence.

This avoids the need for complicated conversations and validation studies be conducted every time technology changes.

The TSP requirements and QPs can be made publicly available for everyone to use to qualify their technologies.

The value of the Stack Model

You might ask: "How is this different from other frameworks, like DiMe V3"?

The Stack Model is actually informed by other foundational frameworks, regulatory requirements and other best practices. It combines them and includes a few additional elements.

The Stack Model is consistent with DiMe V3 and everything in V3 is included in the Stack Model, it is simply organized slightly differently to enable the blocks to be used in an end-to-end workflow. For example, the Clinical Validation information in V3 is associated with the MD layer of the Stack Model to inform clinical interpretation of the measure. Analytical validation may be relevant within both the TSP and DMS blocks.

The stack model also adds a lot of detail, each of the blocks have multiple layers and each of those layers have detailed checklists to make sure everything relevant is covered.

Sound complicated?

It is! Digital measures have many moving parts and a lot of details. Frameworks and guidance is also still evolving. However, what makes this a LOT easier in practice is the DEEP platform. The platform implements the stack model in practice and it guides users through the structure in a user-friendly way.

You don’t need to fully master the theory (though it certainly helps); the system makes applying it in practice much more straightforward. Additionally, the DEEP team consists of experts who can support you in multiple ways - by handling tasks on your behalf, training you to do them yourself or providing assistance in other areas.

The DEEP catalog also has many ready-to-go published 'blocks' in the Public Catalog that you can just access, use and repurpose for your own specific use cases.

Sign up for our newsletter to receive the latest solutions, new opportunities and expert insights on digital health measures delivered to your inbox.

View full post