skip to main content

This content will become publicly available on June 27, 2024

Title: Detecting Sources of Healthcare Associated Infections
Healthcare acquired infections (HAIs) (e.g., Methicillin-resistant Staphylococcus aureus infection) have complex transmission pathways, spreading not just via direct person-to-person contacts, but also via contaminated surfaces. Prior work in mathematical epidemiology has led to a class of models – which we call load sharing models – that provide a discrete-time, stochastic formalization of HAI-spread on temporal contact networks. The focus of this paper is the source detection problem for the load sharing model. The source detection problem has been studied extensively in SEIR type models, but this prior work does not apply to load sharing models.We show that a natural formulation of the source detection problem for the load sharing model is computationally hard, even to approximate. We then present two alternate formulations that are much more tractable. The tractability of our problems depends crucially on the submodularity of the expected number of infections as a function of the source set. Prior techniques for showing submodularity, such as the "live graph" technique are not applicable for the load sharing model and our key technical contribution is to use a more sophisticated "coupling" technique to show the submodularity result. We propose algorithms for our two problem formulations by extending existing algorithmic results from submodular optimization and combining these with an expectation propagation heuristic for the load sharing model that leads to orders-of-magnitude speedup. We present experimental results on temporal contact networks based on fine-grained EMR data from three different hospitals. Our results on synthetic outbreaks on these networks show that our algorithms outperform baselines by up to 5.97 times. Furthermore, case studies based on hospital outbreaks of Clostridioides difficile infection show that our algorithms identify clinically meaningful sources.  more » « less
Award ID(s):
1955939 1955883 2028586 2200269 1918770 1918656
Author(s) / Creator(s):
; ; ; ; ; ; ;
Date Published:
Journal Name:
Proceedings of the AAAI Conference on Artificial Intelligence
Page Range / eLocation ID:
4347 to 4355
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Sexually transmitted diseases (STDs) are detrimental to the health and economic well-being of society. Consequently, predicting outbreaks and identifying effective disease interventions through epidemiological tools, such as compartmental models, is of the utmost importance. Unfortunately, the ordinary differential equation compartmental models attributed to the work of Kermack and McKendrick require a duration of infection that follows the exponential or Erlang distribution, despite the biological invalidity of such assumptions. As these assumptions negatively impact the quality of predictions, alternative approaches are required that capture how the variability in the duration of infection affects the trajectory of disease and the evaluation of disease interventions. So, we apply a new family of ordinary differential equation compartmental models based on the quantity person-days of infection to predict the trajectory of disease. Importantly, this new family of models features non-exponential and non-Erlang duration of infection distributions without requiring more complex integral and integrodifferential equation compartmental model formulations. As proof of concept, we calibrate our model to recent trends of chlamydia incidence in the U.S. and utilize a novel duration of infection distribution that features periodic hazard rates. We then evaluate how increasing STD screening rates alter predictions of incidence and disability adjusted life-years over a five-year horizon. Our findings illustrate that our family of compartmental models provides a better fit to chlamydia incidence trends than traditional compartmental models, based on Akaike information criterion. They also show new asymptomatic and symptomatic infections of chlamydia peak over drastically different time frames and that increasing the annual STD screening rates from 35% to 40%-70% would annually avert 6.1-40.3 incidence while saving 1.68-11.14 disability adjusted life-years per 1000 people. This suggests increasing the STD screening rate in the U.S. would greatly aid in ongoing public health efforts to curtail the rising trends in preventable STDs. 
    more » « less
  2. Abd El-Aty, A. M. (Ed.)
    Background Higher viral loads in SARS-CoV-2 infections may be linked to more rapid spread of emerging variants of concern (VOC). Rapid detection and isolation of cases with highest viral loads, even in pre- or asymptomatic individuals, is essential for the mitigation of community outbreaks. Methods and findings In this study, we analyze Ct values from 1297 SARS-CoV-2 positive patient saliva samples collected at the Clemson University testing lab in upstate South Carolina. Samples were identified as positive using RT-qPCR, and clade information was determined via whole genome sequencing at nearby commercial labs. We also obtained patient-reported information on symptoms and exposures at the time of testing. The lowest Ct values were observed among those infected with Delta (median: 22.61, IQR: 16.72–28.51), followed by Alpha (23.93, 18.36–28.49), Gamma (24.74, 18.84–30.64), and the more historic clade 20G (25.21, 20.50–29.916). There was a statistically significant difference in Ct value between Delta and all other clades (all p.adj<0.01), as well as between Alpha and 20G (p.adj<0.05). Additionally, pre- or asymptomatic patients (n = 1093) showed the same statistical differences between Delta and all other clades (all p.adj<0.01); however, symptomatic patients (n = 167) did not show any significant differences between clades. Our weekly testing strategy ensures that cases are caught earlier in the infection cycle, often before symptoms are present, reducing this sample size in our population. Conclusions COVID-19 variants Alpha and Delta have substantially higher viral loads in saliva compared to more historic clades. This trend is especially observed in individuals who are pre- or asymptomatic, which provides evidence supporting higher transmissibility and more rapid spread of emerging variants. Understanding the viral load of variants spreading within a community can inform public policy and clinical decision making. 
    more » « less
  3. Abstract. Advances in ambient environmental monitoring technologies are enabling concerned communities and citizens to collect data to better understand their local environment and potential exposures. These mobile, low-cost tools make it possible to collect data with increased temporal and spatial resolution, providing data on a large scale with unprecedented levels of detail. This type of data has the potential to empower people to make personal decisions about their exposure and support the development of local strategies for reducing pollution and improving health outcomes. However, calibration of these low-cost instruments has been a challenge. Often, a sensor package is calibrated via field calibration. This involves colocating the sensor package with a high-quality reference instrument for an extended period and then applying machine learning or other model fitting technique such as multiple linear regression to develop a calibration model for converting raw sensor signals to pollutant concentrations. Although this method helps to correct for the effects of ambient conditions (e.g., temperature) and cross sensitivities with nontarget pollutants, there is a growing body of evidence that calibration models can overfit to a given location or set of environmental conditions on account of the incidental correlation between pollutant levels and environmental conditions, including diurnal cycles. As a result, a sensor package trained at a field site may provide less reliable data when moved, or transferred, to a different location. This is a potential concern for applications seeking to perform monitoring away from regulatory monitoring sites, such as personal mobile monitoring or high-resolution monitoring of a neighborhood. We performed experiments confirming that transferability is indeed a problem and show that it can be improved by collecting data from multiple regulatory sites and building a calibration model that leverages data from a more diverse data set. We deployed three sensor packages to each of three sites with reference monitors (nine packages total) and then rotated the sensor packages through the sites over time. Two sites were in San Diego, CA, with a third outside of Bakersfield, CA, offering varying environmental conditions, general air quality composition, and pollutant concentrations. When compared to prior single-site calibration, the multisite approach exhibits better model transferability for a range of modeling approaches. Our experiments also reveal that random forest is especially prone to overfitting and confirm prior results that transfer is a significant source of both bias and standard error. Linear regression, on the other hand, although it exhibits relatively high error, does not degrade much in transfer. Bias dominated in our experiments, suggesting that transferability might be easily increased by detecting and correcting for bias. Also, given that many monitoring applications involve the deployment of many sensor packages based on the same sensing technology, there is an opportunity to leverage the availability of multiple sensors at multiple sites during calibration to lower the cost of training and better tolerate transfer. We contribute a new neural network architecture model termed split-NN that splits the model into two stages, in which the first stage corrects for sensor-to-sensor variation and the second stage uses the combined data of all the sensors to build a model for a single sensor package. The split-NN modeling approach outperforms multiple linear regression, traditional two- and four-layer neural networks, and random forest models. Depending on the training configuration, compared to random forest the split-NN method reduced error 0 %–11 % for NO2 and 6 %–13 % for O3. 
    more » « less
  4. In the early stages of a pandemic, epidemiological knowledge of the disease is limited and no vaccination is available. This poses the problem of determining an Early Mitigation Strategy. Previous studies have tackled this problem through finding globally influential nodes that contribute the most to the spread. These methods are often not practical due to their assumptions that (1) accessing the full contact social network is possible; (2) there is an unlimited budget for the mitigation strategy; (3) healthy individuals can be isolated for indefinite amount of time, which in practice can have serious mental health and economic consequences. In this work, we study the problem of developing an early mitigation strategy from a community perspective and propose a dynamic Community-based Mitigation strategy, ComMit. The distinguishing features of ComMit are: (1) It is agnostic to the dynamics of the spread; (2) does not require prior knowledge of contact network; (3) it works within a limited budget; and (4) it enforces bursts of short-term restriction on small communities instead of long-term isolation of healthy individuals. ComMit relies on updated data from test-trace reports and its strategy evolves over time. We have tested ComMit on several real-world social networks. The results of our experiments show that, within a small budget, ComMit can reduce the peak of infection by 73% and shorten the duration of infection by 90%, even for spreads that would reach a steady state of non-zero infections otherwise (e.g., SIS contagion model). 
    more » « less
  5. Abstract During infectious disease outbreaks, individuals may adopt protective measures like vaccination and physical distancing in response to awareness of disease burden. Prior work showed how feedbacks between epidemic intensity and awareness-based behaviour shape disease dynamics. These models often overlook social divisions, where population subgroups may be disproportionately impacted by a disease and more responsive to the effects of disease within their group. We develop a compartmental model of disease transmission and awareness-based protective behaviour in a population split into two groups to explore the impacts of awareness separation (relatively greater in- vs. out-group awareness of epidemic severity) and mixing separation (relatively greater in- vs. out-group contact rates). Using simulations, we show that groups that are more separated in awareness have smaller differences in mortality. Fatigue (i.e. abandonment of protective measures over time) can drive additional infection waves that can even exceed the size of the initial wave, particularly if uniform awareness drives early protection in one group, leaving that group largely susceptible to future infection. Counterintuitively, vaccine or infection-acquired immunity that is more protective against transmission and mortality may indirectly lead to more infections by reducing perceived risk of infection and therefore vaccine uptake. Awareness-based protective behaviour, including awareness separation, can fundamentally alter disease dynamics. Social media summary: Depending on group division, behaviour based on perceived risk can change epidemic dynamics & produce large later waves. 
    more » « less