CNPEM researchers have shown that machine learning can boost the precision of results from clinical and environmental analysis devices by 70% to 99%, offering a more accessible and pragmatic investment option
One of the nanotechnology research groups at the Brazilian Center for Research in Energy and Materials (CNPEM) has concluded that devices with electrochemical sensors like those used for monitoring environmental data or testing to diagnose diseases can gain an enormous advantage from using computer learning algorithms (popularly known as machine learning). The group’s discussions and conclusions related to this topic were featured on the cover of a special volume of the journal Analytical and Bioanalytical Chemistry directed at young researchers.
The group confirmed that applying artificial intelligence to the results obtained from electrochemical sensors could boost the accuracy of the final data by 70% to even 99% over results from the same device without the use of machine learning. In other words, this method can minimize false positives as well as false negatives in the data from these devices, and consequently can also impact costs.
According to the CNPEM researcher responsible for the study, Renato Sousa Lima, the combination of greater precision and reduced device cost is precisely one of the greatest implications of this practice. He explains that often it is not necessary to invest time and money in sensor improvement, but rather in data processing. “Instead of concentrating research and fabrication efforts on just refining the device itself, which makes it more expensive, we should also look to software that increases the precision of the chemical analyses, in other words, a more accessible option that still ensures reliable results”, states Lima.
The potential applications are varied and limitless. The group that published the study worked with the design, fabrication, characterization, and application of electrochemical sensors in two main areas: environmental analyses and clinical diagnostics. Some examples that have resulted from these studies include “wearable” sensors applied to leaves to monitor physiological conditions in soy and sugarcane plantations and rapid tests for Covid-19.
The use of machine learning is a current topic with a variety of applications. This global trend is growing rapidly, along with scientific studies as a whole. Still, incorporating machine learning into devices like sensors for analyses is still incipient. Lima believes this is why a critical review of this topic has gained clinical relevance, and along with his colleagues he maintains that the use of artificial intelligence will become increasingly common in the field of sensor development.
The study highlighted some machine learning methods that support the final conclusion about the great potential this new option offers. But Lima warns that the solution is only more economical as long as the algorithms chosen for the task do not make the analyses more complex. “It is generally desirable for sensors to generate simple and rapid analyses so they can be operated as part of the daily routine of non-specialists, since there is no point improving the results of a diagnostic if it is difficult to read.” Indeed, today some methods can automatically process data with cell phones, permitting simpler and more connected analysis, as we see in the case of telemedicine. In these cases, says Lima, “there are only benefits!”.
A sophisticated and effervescent environment for research and development, unique in Brazil and present in few scientific centers in the world, the Brazilian Center for Research in Energy and Materials (CNPEM) is a private non-profit organization, under the supervision of the Ministry of Science, Technology and Innovation (MCTI). The Center operates four National Laboratories and is the birthplace of the most complex project in Brazilian science – Sirius – one of the most advanced synchrotron light sources in the world. CNPEM brings together highly specialized multi-thematic teams, globally competitive laboratory infrastructures open to the scientific community, strategic lines of investigation, innovative projects in partnership with the productive sector and training of researchers and students. The Center is an environment driven by the search for solutions with impact in the areas of Health, Energy and Renewable Materials, Agro-environment, and Quantum Technologies. As of 2022, with the support of the Ministry of Education (MEC), CNPEM expanded its activities with the opening of the Ilum School of Science. The interdisciplinary higher course in Science, Technology and Innovation adopts innovative proposals with the aim of offering excellent, free, full-time training with immersion in the CNPEM research environment. Through the CNPEM 360 Platform, it is possible to explore, in a virtual and immersive way, the main environments and activities of the Center, visit: https://pages.cnpem.br/cnpem360/.