Biology subjects
Publications and source records attributed to Santillana, M..
A real-time methodology for monitoring flu activity in middle income countries that is simultaneously accurate and generalizable has not yet been presented. We demonstrate here that a self-correcting machine learning method leveraging Internet-based search activity produces reliable and timely flu estimates in multiple Latin American countries.
BackgroundWidely recognized as a major public health threat globally, the rapid increase of antibiotic resistance in bacteria could soon render our most effective method to combat infections obsolete. Factors influencing the burden of resistance in human populations remain poorly described, though temperature is known to play an important role in mechanisms at the bacterial level.\n\nMethodsWe performed an ecologic analysis of country level antibiotic resistance prevalence in 3 common bacterial pathogens across 28 countries in Europe, and used multivariable models to evaluate associations with minimum temperature and other predictors over a 17-year period (2000-2016). We quantified the effects of minimum temperature, population density, and antibiotic consumption on the rate of change of antibiotic resistance across geographies.\n\nFindingsFor three common bacterial pathogens and four classes of antibiotics, we found evidence of a long-term effect of ambient minimum temperature on rates of increase of antibiotic resistance across 28 countries in Europe between 2000-2016. Specifically, we show that across all antibiotic classes for the pathogens E. coli and K. pneumoniae, European countries with 10{degrees}C warmer ambient temperatures have experienced more rapid increases in antibiotic resistance over the 17-year period, ranging between 0.33%/year (95% CI 0.2, 0.5) and 1.2%/year (0.4, 1.9), even after accounting for recognized drivers of resistance including antibiotic consumption and population density. We found a decreasing relationship for S. aureus and methicillin of -0.4%/year (95% CI -0.7, 0.0), reflecting widespread declines in MRSA across Europe over the study period.\n\nInterpretationAmbient temperature may be an important modulator of the rate of change of antibiotic resistance. Our findings suggest that rising temperatures globally may hasten the spread of resistance and complicate efforts to mitigate it.\n\nFundingCanadian Institutes of Health Research Fellowship (D.R.M.).
In the presence of population-level health threats, precision public health approaches seek to provide the right intervention to the right population at the right time. Accurate real-time surveillance methodologies that can estimate infectious disease activity ahead of official healthcare-based reports, in relevant spatial resolutions, are critical to eventually achieve this goal. We introduce a novel methodological framework for this task which dynamically combines two distinct flu tracking techniques, using ensemble machine learning approaches, to achieve improved flu activity estimates at the state level in the US. The two predictive techniques behind the proposed ensemble methodology, named ARGONet, utilize (1) a dynamic and self-correcting statistical approach to combine flu-related Google search frequencies, information from electronic health records, and historical trends within a given state, as well as (2) a data-driven network-based approach that leverages spatial and temporal synchronicities observed in historical flu activity across states to improve state-level flu activity estimates. The proposed ensemble approach considerably outperforms each individual method and any previously proposed state-specific method for flu tracking, with higher correlations and lower prediction errors.