The public health system in Chile covers 75% of the population, and while it is efficient by many health indicators, it suffers from long waiting times, high patient no-show rates, and relatively slow adoption of key developments from fields such as genomics, operations research, and data science. The COVID-19 pandemic has also highlighted many other issues that could benefit from a modeling and data analytics perspective. Over the past ten years, several CMM teams have invested a great deal of effort to work on health problems, using diverse and complementary mathematical techniques. Because of the above, a new line of research called Digital Health has been created that unifies and drives these efforts.
We identified the following topics of interest:
We plan to collaborate with relevant stakeholders seeking to improve access and delivery of healthcare in the public system. An example is the work to improve the management of hospital schedules, through the collaboration agreements we have signed with three major public health institutions in Chile and the funding we have secured for a two-year project on building a software prototype to address no-show. Another challenge is the Chilean organ donation system, which is undergoing a major overhaul. Addressing the organ allocation problem efficiently and fairly requires several techniques from different areas, such as random graph models, combinatorial optimization techniques and data analysis.
In biomedicine we focus on the increasingly common use of imaging techniques to support medical diagnosis and biological research. We work on improving these techniques in cardiac and pulmonary research using appropriate biophysical models, seeking a better understanding of the mathematics used in light sheet microscopy techniques, and on mathematical contributions to the development of new techniques for the early detection of bone weakness from ultrasound measurements.
More information: Numerical Biomedicine website
Clinical natural language processing
In clinical natural language processing, we are focused on the creation of linguistic resources, particularly the first annotated corpus using Chilean narratives. Besides, we work on proposing neural network-based models to extract key information (named entity recognition). Our case study is the non-GES Waiting List in Chilean public hospitals, with the ultimate goal of supporting clinical decision-making and automatizing the secondary use of information. Annotation guidelines, word embeddings, annotated text, and a code repository can be found on our webpage.
More information: Natural Language Processing website
On the other hand, based on the group’s experience in modeling infectious diseases (dengue and STD, among others), as well as their impact on social behavior (N1H1), last year we created a working group on COVID-19 in Chile. We have estimated the impact of the application of different non-pharmacological mitigation actions in Chile, in particular on the demand for hospital beds (our periodic reports are available at https://covid-19.cmm.uchile.cl/). Members of the CMM also participate in the government data board on COVID-19. We plan to continue working in this line, especially in the development of cost-effective approaches in decision making, aimed at prevention and treatment and in the correct evaluation of their social impacts.
Human health genomics
In the field of human health genomics, new sequencing technologies, together with our improvements in genome assembly algorithms using hybrid libraries, have made it possible to obtain genomes from almost any species at very low cost. In particular, today we are competitive in developing de novo methods to assemble human and other more complex genomes based on short sequences at a cost of $5,000 or less. This opens up opportunities, such as functional tracking of cancer cells over time. There are advanced discussions on this idea with the Arturo Lopez Perez Foundation, one of the most important private cancer institutions in Chile.
More information: Mathomics website