Search

Understanding the effect of media on disease spread can help improve epidemic forecasting and uncover preventive measures to slow the spread of disease. Most previously introduced models have approximated media effect through disease incidence, making media influence dependent on the size of epidemic. We propose an alternative approach, which relies on real data about disease coverage in the news, allowing us to model low incidence/high interest diseases, such as SARS, Ebola or H1N1. We introduce a network-based model, in which disease is transmitted through local interactions between individuals and the probability of transmission is affected by media coverage. We assume that media attention increases self-protection (e.g. hand washing and compliance with social distancing), which, in turn, decreases disease model. We apply the model to the case of H1N1 transmission in Mexico City in 2009 and show how media influence—measured by the time series of the weekly count of news articles published on the outbreak—helps to explain the observed transmission dynamics. We show that incorporating the media attention based on the observed media coverage of the outbreak better estimates the disease dynamics from what would be predicted by using media function that approximate the media impact using the number of cases and rate of spread. Finally, we apply the model to a typical influenza season in Washington, DC and estimate how the transmission pattern would have changed given different levels of media coverage. [ABSTRACT FROM AUTHOR]

Background: Studies of cost-effective disease prevention have typically focused on the tradeoff between the cost of disease transmission and the cost of applying control measures. We present a novel approach that also accounts for the cost of social disruptions resulting from the spread of disease. These disruptions, which we call social response, can include heightened anxiety, strain on healthcare infrastructure, economic losses, or violence. Methodology: The spread of disease and social response are simulated under several different intervention strategies. The modeled social response depends upon the perceived risk of the disease, the extent of disease spread, and the media involvement. Using Monte Carlo simulation, we estimate the total number of infections and total social response for each strategy. We then identify the strategy that minimizes the expected total cost of the disease, which includes the cost of the disease itself, the cost of control measures, and the cost of social response. Conclusions: The model-based simulations suggest that the least-cost disease control strategy depends upon the perceived risk of the disease, as well as media intervention. The most cost-effective solution for diseases with low perceived risk was to implement moderate control measures. For diseases with higher perceived severity, such as SARS or Ebola, the most cost-effective strategy shifted toward intervening earlier in the outbreak, with greater resources. When intervention elicited increased media involvement, it remained important to control high severity diseases quickly. For moderate severity diseases, however, it became most cost-effective to implement no intervention and allow the disease to run its course. Our simulation results imply that, when diseases are perceived as severe, the costs of social response have a significant influence on selecting the most cost-effective strategy. [ABSTRACT FROM AUTHOR]

In many data classification problems, a number of methods will give similar accuracy. However, when working with people who are not experts in data science such as doctors, lawyers, and judges among others, finding interpretable algorithms can be a critical success factor. Practitioners have a deep understanding of the individual input variables but far less insight into how they interact with each other. For example, there may be ranges of an input variable for which the observed outcome is significantly more or less likely. This paper describes an algorithm for automatic detection of such thresholds, called the Univariate Flagging Algorithm (UFA). The algorithm searches for a separation that optimizes the difference between separated areas while obtaining a high level of support. We evaluate its performance using six sample datasets and demonstrate that thresholds identified by the algorithm align well with published results and known physiological boundaries. We also introduce two classification approaches that use UFA and show that the performance attained on unseen test data is comparable to or better than traditional classifiers when confidence intervals are considered. We identify conditions under which UFA performs well, including applications with large amounts of missing or noisy data, applications with a large number of inputs relative to observations, and applications where incidence of the target is low. We argue that ease of explanation of the results, robustness to missing data and noise, and detection of low incidence adverse outcomes are desirable features for clinical applications that can be achieved with relatively simple classifier, like UFA. [ABSTRACT FROM AUTHOR]