Research

In Chile, 74% of the population is covered by the governmental health funds, and in order to see an specialist, patients have to be referred by a primary care physician, entering in this way to a waiting list. The government introduced the GES plan, which groups a series of specialist consultations and surgeries that are prioritized, creating a huge difference in waiting times by diseases covered and not by GES. Check the paper where we reported an increased mortality within the non-GES group and the waiting list visualization, a joint project with Fabián Villena (paper that explains the weighted keywords was published in Rev Medica Chile). The annotation of this referrals to train Named Entity Recognition is explain in our EMNLP long paper.

This project proposes the development of a framework to create, test and apply vector representations of clinical text produced in Chile. The proposed technique, neural word embeddings, has shown good performance in general texts as well as in biomedical applications. Testing the quality of embeddings within the medical context and in Spanish language is not trivial, and it is an important component of this project. Joint project with Fabián Villena and Jorge Pérez.

Using a variety of machine learning methods we did the exercise of predicting the obesity prevalence in 79 countries based on food sales in 48 food and beverage categories. In the picture is shown one of the trees used by Random Forest, which shown the best performance among the tested methods. We found that the most relevant food category to predict obesity is baked goods and flours, followed by cheese and carbonated drinks. Work done in collaboration with Felipe Tobar and others, and the paper was published by the Health Informatics Journal.

In October 2014, Chile implemented a tax modification on SSBs called the Impuesto Adicional a las Bebidas Analcohólicas. In collaboration with Cristóbal Cuadrado and Nicolás Silva from University of Chile and Andrew Mirelman, Ryota Nakamura and Marc Suhrcke from York University we have worked in estimating the effect of this tax. Check our papers in PLOS Medicine and Social Science & Medicine.

This was my PhD thesis. In this paper the experiments, theory and simulations I performed to understand the active convention in the absence of motility is explained.