Artificial intelligence will help diagnose patients

28. 06. 2022

Artificial intelligence will help with diagnosing patients on the basis of their lab samples. The unique pilot technology called MAIA is the joint project of the Biology Centre of the CAS, České Budějovice Hospital, and biotech company Aiolite. The project is co-financed by the Technology Agency of the Czech Republic, and it culminated at the end of June this year. The collaboration will result in the development of an advanced software tool that will help doctors make informed decisions about patients' further treatment.

The aim of the MAIA (Metabolomic Artificial Intelligence Analysis) project was to link data obtained from patients' urine and blood samples and develop an efficient tool for their analysis and synthesis using artificial intelligence methods. The results will be used by physicians to monitor the development of the patient's health and to choose further treatment.

The MAIA project is a successful example of linking research results with clinical medicine and industry. It involves three teams of experts – physicians, researchers, and computer scientists – led by young executives under the supervision of mentors.

For the pilot project, doctors from České Budějovice Hospital selected patients from the Intensive Care Unit of the Infectious Diseases Department with suspected septic complications. From standard biological material collections, anonymised samples processed at -20 degrees are sent to the biochemistry laboratory of the CAS Biology Centre. At the same time, anonymised clinical data on the patient's condition and treatment, as well as laboratory results related to the time and course of the disease, are sent with the samples to the MAIA repository.

Biochemists from the Biology Centre perform so-called metabolomic analysis on the samples. "We use the latest findings and methods from metabolism research. In a tiny amount of a human blood serum or urine sample, we can simultaneously measure about an order of magnitude more substances than are usually obtained in standard clinical analyses nowadays," says Petr Šimek, head of the Laboratory of Analytical Biochemistry and Metabolomics at the Biology Centre.

Artificial intelligence in practice

The huge data sets obtained are then analysed, sorted, and evaluated by the MAIA software tool with artificial intelligence algorithms from Aiolite. "This so-called big data contains information on a wide range of potential biomarkers and the medication used, which is then evaluated by our AI-equipped server. The treating physician receives comprehensive information to assess the patient's current health state, which helps them make decisions on a qualitatively higher level about the optimal course of treatment. This tool can thus improve the patient's quality of life or even save their life," explains Jakub Schůrek, CEO of Aiolite.

The functionality of the MAIA platform is also being continuously checked by a parallel manual processing using conventional methods. The final evaluation of the analysis results was carried out by a team of doctors from the infectious diseases department, clinical biochemists, and other experts from laboratory medicine. The aim of the project was to develop pilot tools for complex integration, processing, and use of all available clinical data with the intention of searching for new, previously unrecognised relationships and information and applying them to clinical diagnostics.

It will help physicians assess the course of treatment

The researchers first tested the new approaches in selected cell culture models whose physiological state can be better controlled. Then, they proceeded to work on primary lymphocytes, and finally on samples taken from patients in a septic state.

"The project caught our attention by using a wide range of determinable parameters of metabolic transformation in the human body, which has the potential to assess the energy status of the organism, the ability to produce the necessary substances and to evaluate the effect of the treatment measures. Due to the number of measurable parameters, their dynamics, the influence of biorhythms, the treatment regime, and other factors, it is very difficult to find patterns usable for assessing liminal situations suitable for determining further progress or a fundamental change of strategy, without the use of artificial intelligence elements," says Miroslav Verner, Director of the Division of Central Laboratories of České Budějovice Hospital.

Thanks to cutting-edge analytical technology and the possibilities of artificial intelligence, Verner adds, this should lead to developing a proven tool to help healthcare professionals detect decision-making breaking points in time in order to reduce the number of deaths from sepsis and also to render treatment and recovery time more efficient and shorter.

The output of the MAIA pilot project will be an optimised setting up of processes and precise procedures for data evaluation using artificial intelligence, which will allow the project to move to the next phase of validating the system's functionality on a larger database of patients.

The joint project will be continued. "The collaboration with my colleagues is absolutely perfect. We have created a functional team of experts and specialists with ongoing ambition to work on projects that will enable the linking of research findings with commercial practice. We are finalising the MAIA II project, which will take the findings and conclusions from the first project to the next stage of validation and possible applications," adds Jakub Schůrek.

Artificial intelligence will help diagnose patients

Read also

Contacts for Media