A web application, used by laboratories worldwide.
In hospital laboratories, machines perform automated analyses. By means of the measured values, real-time calculations can be performed in order to see whether the analysis process still functions properly. For Huvaros, Iquality developed a web application that laboratories can use to analyse how they can setup their automated analysis machine in such a way that it quickly raises the alarm if something may be wrong, but not unnecessarily often.
Better settings for analysis machines in laboratories
Huvaros is a medical startup, founded by Huub van Rossum. As a clinical chemist, he is responsible for analysis machines that perform thousands of automated analyses every day. Human hands are barely required anymore. Therefore, it's important that the machines raise the alarm when they measure values that could point to technical problems or errors in the uptake or preparations. For this purpose, the machines perform calculations for their own measurement results.
"For every analysis machine, the values that require the sounding of the alarm are different," Huub explains: "The type of research, type of hospital, period in the week or year, all these factors influence the possible values of your measurements. Of course, it's important that the alarm is raised on time, so no incorrect analyses sneak through. At the same time, the alarm should not be raised unnecessarily, as this would mean a new investigation of the potential problem every single time. As you can see, a small difference in your settings could mean a lot of extra work."
From trail-and-error to simulation
At present, laboratories set their machines via trial-and-error. If a machine gives too many unnecessary alarms, the lab technicians try to devise which settings would be better for that machine. They enter them and a week or month later, they check whether the number of unnecessary alarms has actually decreased. Huub sought to improve this method and invented a way of conducting simulations on the basis of the measurement data of a machine plus added incorrect data. The results of this provide valuable insight into what the most suitable settings are for a machine within a certain environment and type of investigation.
There are so many variables that influence the measurements. Therefore, it is very important that the alarm is sounded in time when something is indeed wrong.Drs. Huub van Rossum, Owner Huvaros
Worldwide useable web application
Huub van Rossum had initially designed his tool in Excel, but the possibilities of this were limited. Therefore, he asked Iquality to develop a web application. Huub: "Iquality has made a web application that can conduct thousands of simulations within a few minutes, all on the basis of measurement data a machine has generated itself. By means of this, it can be accurately determine which settings are needed for such a machine to raise the alarm in time, but not unnecessarily so. This eventually means that more analyses with a guaranteed quality can be performed every day."
Working with a fixed budget
For the development of the web application, Huub has opted for the concept of Team as a Service (TaaS). This means that the project has been conducted with a multidisciplinary team that is scalable and can, in principle, perform everything a customer needs. Moreover, the members of such a team are well coordinated, making them extra efficient. Huub is very satisfied about the approach: “I had a fixed budget. For that amount, I wanted to develop a functional web application that could also be used by other laboratories. This means that I had a number of minimum requirements, but also that we left lots of details open beforehand. We just got started and looked at what we could perform every week."
A good start
The web application of Huvaros can be used by every laboratory in the world. Van Rossum has already sold the first licenses. If there is sufficient interest, he wants to further develop the web application: "The simulation now works on the basis of two often used algorithms, but there are also analysis machines that use different algorithms. I want to add these in future. Just like extra parameters for the simulations. Furthermore, user management now takes place via Iquality. If the number of users increases, a special module may be more useful for this purpose."
When the Huvaros web application gets more users, the principle of machine learning could also become interesting. For example, users could see what choices other users have made in similar situations. The concept for this was already investigated during a session of Iquality LABS where specialists of Iquality, Huub van Rossum and a student applied mathematics investigated the possibilities of machine learning for the Huvaros web application for a day.