06Mar2026
14:00 Master's Defense room 85 of IC2
Topic on
Unsupervised machine learning in discovering evolutionary profiles of COVID-19 in laboratory tests.
Student
Gabriel de Freitas Leite
Advisor / Teacher
André Santanchè - Co-supervisor: Susan Elisabeth Domingues Costa Jorge
Brief summary
The COVID-19 pandemic has proven to be an unprecedented health crisis, triggering challenges to health systems globally. In this context, the long-term impacts of this pandemic on the population health profile remain an open question. A notable example is "long COVID," a condition in which affected individuals do not recover for weeks or months after the initial infection. Therefore, there is a need to study the evolution of the population health profile as a consequence of the pandemic. This work aims to develop a method based on clustering techniques to group clinical profiles of patients impacted by COVID and their clinical evolution. The analysis and discovery of profiles, however, are profoundly affected by the choice of computational clustering technique used. Each algorithm has distinct premises, adapting better to different data structures, such as dimensionality, density, or the expected geometry of the groups. To identify late effects, the study focuses on how the health profile changed after COVID through a comparative analysis. The change is characterized through a contrastive analysis of the patient's trajectory. The method contrasts the clinical profiles found at the first COVID admission with the characteristics acquired in subsequent admissions, thus allowing the identification of sequelae patterns according to the original profile. It is important to note that, to date, there is no comparative study evaluating the impact of different clustering techniques in the specific context of the evolution of COVID clinical profiles. This study will contribute to characterizing how each technique performs in this type of study and what the selection criteria are according to the objective. The results in the health context have the potential to identify relevant correlations, promoting greater knowledge about the sequelae of COVID.
Examination Board
Headlines:
| André Santanchè | IC / UNICAMP |
| Gustavo Jacob Lourenço | FCM / UNICAMP |
| Marcelo da Silva Reis | IC / UNICAMP |
Substitutes:
| Guilherme Pimentel Telles | IC / UNICAMP |
| Paula Dornhofer Paro Costa | FEEC/UNICAMP |