In medicine, the procedure of drawing blood from a patient for laboratory tests is called phlebotomy. Inpatients are submitted to phlebotomy protocols that include successive requests for tests, at rates that can reach daily extractions, in order to diagnose or monitor the evolution of their clinical conditions. However, frequent extractions expose the patient to high blood loss in a short period of time, which can trigger health problems, since the time between extractions can be insufficient for the adequate replacement of lost blood components. In order to evaluate the hypothesis of phlebotomy blood loss to be a relevant factor for the development or worsening of anemia and for an increase in hospital length of stay, this research conducted analyzes guided by retrospective clinical, laboratory and demographic data of 28,312 hospitalizations that occurred in the Hospital of Clinics at Unicamp (HC/Unicamp) between the years 2012 and 2016. In an initial phase, exploratory investigations were conducted in order to capture the main characteristics inherent to the dataset. Then, an interpretable Machine Learning (ML) approach was developed, using the Decision Tree-based techniques Gradient Boosting Machines and Random Forests for regression, binary classification and multiclass classification tasks. In order for the computational solutions produced to offer a degree of interpretability that is convenient for use in the medical-hospital setting, methods to extract interpretability of ML models were used to express the relationship between the different variables used, as well as listing the degree of their contributions to the models outputs. The results suggest a relationship between the amount of blood samples taken during hospitalization and length of stay, as well as the decrease in patients hemoglobin levels, a factor directly related to the development of anemia. The study conducted with unpublished data, to the best of our knowledge, is the first to jointly investigate relationships between phlebotomy, anemia and length of stay using ML. In addition, it deepens on existing knowledge about the problem and expands the correlated literature, since it takes into account the variables adjacent to the possibility of developing anemia in the course of hospitalization, such as blood transfusions and surgeries performed. It also offers HC/Unicamp subsidies for evaluating its phlebotomy processes and grounds for decision making.


Master’s pre-defense

Candidate: Flávia Érika Almeida Giló Azevedo

Advisor: André Santanchè


When: August 7, 2020 at 1 pm