Artificial intelligence to improve lung cancer treatments

Lung cancer is the leading cause of cancer death. According to some estimates, in 2022 there will be more than 236,000 new diagnosed cases and more than 130,000 deaths from this disease. Although these figures have been decreasing in recent years, it is still necessary to improve detection methods and treatments to continue increasing both survival and quality of life for patients.

And it is that current cancer treatments are characterized by the toxicities and side effects they generate. For this reason, finding effective treatments that increase patient survival and reduce toxicities is a primary objective.

In this context, a study carried out by researchers from the Polytechnic University of Madrid (UPM) has focused on the analysis of clinical data on lung cancer patients in order to obtain statistical patterns of the best treatments, taking into account the profile of the disease of the patient being treated.

The study, carried out by researchers from the Medical Data Analysis Laboratory (MEDAL) – Center for Biomedical Technology, and the Department of Computer Languages ​​and Systems and Software Engineering (ETS de Ingenieros Informático) of the Polytechnic University of Madrid, has with the collaboration of the Puerta de Hierro University Hospital, which has provided clinical data on lung cancer treatments.

The research is part of the European project P4-LUCAT, which aims to try to find strategies related to treatments that, on the one hand, can improve patient survival, and on the other, can reduce the toxicities associated with the latter . All this through the integration and analysis of different data sources: electronic medical records, scientific publications and open data (open data).

The figure shows the different unstructured data sources consulted in the P4-LUCAT project to convert this information into structured data. (Image: UPM)

“Artificial intelligence algorithms are applied to these data to try to respond to the objectives set,” explains Alejandro Rodríguez González, professor at the UPM and coordinator of the project. In addition, the project makes use of other technologies, such as those related to natural language processing, to identify relevant terms and associations on data such as symptoms, treatments and effects.

The study presents the preliminary results on the discovery of statistical patterns in lung cancer treatments, taking into account the characteristics of the disease developed in the patient. These patterns have been contrasted with clinical guidelines, documents that include guidelines to be used by oncologists in their daily medical practice.

The results, indicates the UPM researcher, “showed that the patterns found in the treatments prescribed to the patients coincide with the guidelines of the oncology clinical guidelines”. In addition, the authors of the study have detected the need to include in the analyzes information related to the results and side effects of the treatments, in order to carry out a more informed evaluation of the statistical patterns found in the treatments and their comparison with the clinical guidelines.

“The results obtained in the research constitute a first step for the analysis of lung cancer treatment data, and that the results of these analyzes can be integrated into an information system,” says Rodríguez. Said information system may be applied in the clinical environment to help the doctor by providing information about the result that a treatment may have, statistically, depending on the characteristics of the patient being treated. In the future, “these results could have a great impact in the field of health, improving the survival and quality of life of patients, with the consequent reduction in the expenses of the health system”, concludes Guillermo Vigueras, professor at the UPM and technical coordinator of the project. (Source: UPM)