With increasing process complexity in developing Point of Care diagnostic devices, manufacturers face a challenge realizing economic benefits due to uncertain development costs and timelines, instrument costs and device fleet maintenance.

“The rise of connected devices, centralized data storage and machine learning are changing the way POC diagnostics deliver value.”


Guide: Best Practices for Point of Care Product Development


The rise of connected devices, centralized data storage and machine learning (ML), however, are changing the way Point of Care (POC) diagnostics deliver value, and provide a pathway to improved overall efficiency.

The medical device sector has been somewhat slow in adopting some of this new technology, but awareness and uptake are increasing rapidly. Innovative technologies that utilize algorithms and ML now enable Point of Care testing (POCT) devices to produce quantitative lab-quality test results, including some with CLIA waivers. According to a recent article by Dr. Eric Topol, the number of FDA approved uses of artificial intelligence (AI) in healthcare doubled in 2019.1

In this article, we consider the critical success factors in the development and deployment of POCT diagnostic devices – timeline, cost, differentiation – and how the adoption of cloud solutions can impact these factors as a product is defined, developed and deployed.

Leverage cloud software to shorten development timelines

Time is money, and in the context of POC product development, software effort often consumes a significant portion of the development budget. Wherever possible, common software components should be leveraged (e.g., hardware interfaces); however, new regulations and user requirements invariably add new functionality and, in turn, development time.

Over the last decade, the rapidly increasing size of the digital economy in the US (6.9%, or $1,351.3 billion, of current-dollar GDP in 20172) has created a new ecosystem of cloud software that is now entrenched in every aspect of daily lives – in email, games, media, navigation, shopping, banking and more. The corresponding tools, expertise and platforms such as Amazon Web Services, Microsoft Azure and Google Cloud have created a powerful set of software solutions that should be considered as an integral aspect of any POCT device development.

Leveraging solutions such as authentication, encrypted data storage and machine learning outside the device can greatly reduce the software development effort while offering powerful benefits to both the device end-user and the manufacturer.

Separating the critical regulated assay workflow software on the device from the other data interactions (for both patient and device data) allows a more flexible architecture as well as regulatory benefits. Doing so provides a flexible model to define software units with different safety classifications and the opportunity to evolve the cloud software more rapidly within the regulatory guidelines.

Without the constraints of a device software platform, cloud solutions can be easily scaled and globally deployed, offering a rich environment of applications for data reporting, analyses and archiving.

Minimize device footprint and cost through cloud solutions

For any diagnostic product, there is great sensitivity to instrument and consumable manufacturing cost. This is particularly critical for POC product development, as their utilization is typically lower compared to laboratory systems, and there is a need to amortize instrument cost over fewer tests. Furthermore, POCT device purchasing budgets are limited as they are often purchased for walk-in clinics or home use, and their service costs must be contained.

The design of POCT device hardware must be scrutinized to avoid expenses associated with computing resources, screens, keyboards, etc. It’s essential to select a minimum hardware configuration that is focused on executing the assay workflow, while avoiding extra features that are not critical to the diagnostic test.

cloud enabled poct device
Leveraging cloud solutions helps to reduce the footprint and cost of POCT devices.

Most other user interactions such as data storage and access, configurations, data logging and algorithms can easily be moved to the cloud. This enables the reduction of the footprint and cost of the device. Extreme examples of devices with greatly reduced footprint can be seen on wearable devices such as the Apple Watch.

Differentiate diagnostic devices through AI and ML

Even though we may not be fully aware, AI through ML has already permeated most of our lives. Cloud companies have built applications based on email (Google), shopping (Amazon) and social interactions (Facebook, Instagram) that seem to know a lot more about our personal lives than we may be comfortable with.

“AI and ML have the capability to transform the way assay results are gathered and analyzed, speeding up the process and increasing the accuracy of test results.”

In POCT devices, AI and ML have the capability to transform the way assay results are gathered and analyzed, speeding up the process and increasing the accuracy of test results. Utilizing AI and ML is an essential way for product developers to differentiate and add value to their product.

Medical and diagnostic devices typically record device activity (user inputs, hardware activities, data exchanges, events and errors) in a log file. These data can be useful for debugging and servicing issues with the device. However, without a common cloud repository, the full value of this data is not being captured and retained.

Instead of just capturing basic functionality data, envision a database containing a record of every single action on all devices across all hardware and software versions. With ML, prognostic models can be built from these data to enable more effective service, maintenance, user support and marketing analyses.

To give an example from a recent project developing a product with a diagnostic imaging system, Invetech adopted a cloud infrastructure early in the architecture design of the system. With a central image repository, we were able to efficiently annotate the images for the ML modeling task.

As more instruments joined the fleet, the variety of analyzed specimens increased and allowed further expansion of the assay menu simply through iterating the modeling cycle. Based on early comparisons, diagnostic accuracy of the model exceeds the capability of a typical microscope lab operator.

Closing thoughts

Adoption of cloud technologies as an integral part of device development from the start of the project provides numerous advantages for both the development and total lifecycle management of the product.

Technology and customer requirements evolve continuously. Consequently, future Point of Care solutions will invariably include new architectures and features that leverage cloud solutions. There are technical and organizational adoption hurdles that need to be considered, especially cybersecurity and data protection regulations, however, with the right expertise, cloud computing can greatly increase product and operational capabilities while reducing development and product costs.

The list of approved/cleared AI platforms is expanding quickly; now is the time to assess how your product strategy can benefit from this technology and leverage the innovation possibilities that the competition will envy.

Guide: Best Practices for Point of Care Product Development

For more strategies on how to accelerate timelines and decrease costs, download our Best Practices for POC Product Development Guide.


  1. Topol, E.J. High-performance medicine: the convergence of human and artificial intelligence. Nat Med 25, 44–56 (2019) doi:10.1038/s41591-018-0300-7. Published January 7, 2019. Accessed January 2020.
  2. Kevin Barefoot, Dave Curtis, William A. Jolliff, Jessica R. Nicholson, and Robert Omohundro. Research Spotlight Measuring the Digital Economy. Bureau of Economic Analysis, Volume 99, Number 5. Published May 2019. Accessed January 2020.