## Background
## The Problem
The first generation of the medical device had fixed BOM. It was equipped with an industry-grade HMI panel with a touch screen. In addition, the number of ports was limited. The computing power was enough to run the basic operations but was not enough to extend the functionality and use neural networks on the device. The task was to squeeze as much as possible into the current device.
Furthermore, the company’s roadmap highlighted a need to implement the upcoming features. Consequently, the current hardware platform would have been a bottleneck for the project.
## The Solution
We proposed a hardware change, and the Client gave us a green light. After getting all the needed parts, we created a prototype in less than a week. The prototype used a popular, low-cost single-board computer. After completion, we run a POC with highly satisfying results. The performance was a blast. Therefore, the Client accepted our proposal.
It is worth mentioning that the improved device served as a prototyping platform for two consecutive years, clinical study tests were run on it, and it is still being used.
## The Process
We kicked off by understanding the Client’s needs, plans, and vision. Afterward, we communicated with the on-site teams to establish communication processes. During the whole project, the team members were working remotely. However, in the case of milestone boot camps, the team rushed to the Client’s office.
Moreover, we held weekly Teams meetings with the other teams to establish our priorities and allocate resources. Additionally, we kept in touch with the CEO and consulted with him about our next steps to ensure that we were aligned with the company’s guidelines on time and within budget.
What is more, we also took part in preparing documents for regulatory purposes. In consequence, the software development processes had to fit the strict medical regulations.
## The Effect
- main computing unit cost reduction by a factor of 8;
- shortening the software update process by 80% (from 30 to a few minutes);
- boosting computing power enough to run Neural Nets (including Machine Learning models), xgboost, and all the legacy algorithms;
- increasing maximum sampling frequency 1.5x which improved device resolution;
- a possibility to replicate the setup quickly, as a result, the devices without the expensive laser were built to run UI testing and rapid prototyping;
- an update of the base OS allowing to use of the latest frameworks for Machine Learning;
- the hardware fitting in the existing casting allowed for plugging a high-DPI touch screen;
- increasing the number of available ports and GPIOs which allowed the rapid prototyping and extension of the device capabilities;
- improved user experience.
## D-Base Project Resources
Recommendations
About the D-Base project
## Project Background
Although medicine has made astonishing progress in the last century, we are still far from curing a plethora of diseases or making some medical procedures, like blood tests, comfortable.
With this in mind, the DiaMonTech company developed D-Base – a device that painlessly measures blood sugar levels for patients with diabetes. Our role was to provide appropriate software and advise on some hardware matters to achieve the Client’s long-term goals.
## The Team
The project was handled by a 5-person team. With diverse skills encompassing both hardware and software, we were able to deliver the code under the exacting standards for medical devices.
## The Result
Our proposed changes brought the breakthrough the Client was looking for. The single-board computer resolved many problems as well as upcoming challenges making the device faster, flexible and cost-efficient.
