In the context of turbulent world agricultural markets, high food prices, and an estimated 800 million people suffering from chronic hunger, agricultural statistics are critical for economic and social development. This session highlights the importance of accurate and timely agricultural statistics for policymakers, helping increase food production, allocating scarce resources, and adapting to extreme weather situations. Speakers also address the use of newer technologies, such as remote sensing and drones, to monitor agricultural production and food security, and modern approaches to data collection in developing countries.
Section U Business Meeting
It is commonly believed that screening tests for diseases such as cancer result in better outcomes (i.e., reduced mortality, less aggressive treatment regimens, extended survival). Mammography, colonoscopy, and prostate-specific antigen tests are recommended to detect breast, colorectal, and prostate cancers earlier than clinical symptoms would ordinarily appear. However, as with any treatment protocol, the benefits of screening tests should be evaluated. Randomized screening trials provide the most scientifically credible evaluations of the effectiveness of screening programs for early detection: study participants are offered periodic screening, and those in the control group are told to follow their usual medical care. Evaluating disease-specific mortality rates and other data from screening trials is challenging: trials cannot be blind, and they have contamination from study participants who miss screening exams, control participants who get screened, and screening exams with false-positive and false-negative results. Speakers in this session describe practical issues in evaluating screening, the challenges of conducting large randomized screening trials, new developments in statistical analysis of data from screening trials, and estimating error rates in cancer modalities. They discuss National Cancer Institute trials, the European Randomized Study of Screening for Prostate Cancer, and the Canadian National Breast Cancer Screening Study and implications for public policy and national and global recommendations for cancer screening.
Achieving the goals of advancing the practice of science, and specifically increasing the value of science to society, should include raising the level of ethical behavior across the sciences. Science can support the translation of knowledge into viable policy options, but only when that science is conducted and reported in a manner that explicitly supports and respects cultural values. Speakers representing scientific disciplines with differing levels of explicit ethical frameworks will discuss the role of ethics and professional practices within their disciplinary cultures. The discussion also will include how discipline-specific ethics and a culture of professionalism can help to continually improve scientific methods, mechanisms, and outputs, and increase the value of science to society.
As a result of several factors, including a 2009 National Academies report, the role of statistics in forensic science and criminology has become a prominent issue. Many municipalities are exploring the use of predictive policing, in which statistical algorithms are used to forecast the locations and times when criminal activity is most likely to occur; however, the accuracy and value of such tools is still an open question. Similarly, methods from statistical epidemiology are being used to identify regions where drug abuse, rape, or murders tend to occur (and regions in which such crimes are successfully prosecuted). An epidemiological approach can identify common factors that may enable better causal models and more effective intervention. Additionally, several recent shootings by police, caught on camera, have led to the creation of datasets on police homicides. These data are being mined from a number of perspectives, both to improve law enforcement practices and to evaluate the role of race and other factors in police and public interactions. This session provides an overview of different kinds of statistical research and potential uses in addressing the pressing needs of modern criminology.
The Zika epidemic has been declared by the World Health Organization to be a “public health emergency of international concern,” and dengue continues to be a global public health menace. Both viruses are spread by the same Aedes mosquito vector, which now exposes nearly half of the world population to these viruses. In addition, Zika can be spread sexually from male-to-female and possibly through other types of direct contact. These epidemics can be controlled through improved vector control and new vaccines. This session describes the statistical and mathematical methods being developed and used to evaluate the effectiveness of these interventions. Based on their potential levels of effectiveness, the panelists will discuss how these interventions may be used to control Zika, dengue, and other arboviruses spread by Aedes now and in the future.
New and emerging technologies generate low-cost, high-throughput assessments of molecules at the DNA, RNA, and protein levels, with streams of patient data from mobile health monitors. As yet unknown relationships between genomics, symptomatic disease, health interventions, and health outcomes reside in these huge, unstructured data archives. Multidisciplinary quantitative science combining biostatistics, bioinformatics, and computing is needed to reveal new insights for human biology, clinical medicine, and public health. Key goals include identifying mutations that initiate disease and molecular markers that predict patient response to therapy. Finding biomarkers to test in clinical trials is a step toward personalized risk modeling, personalized cancer prevention, and precision medicine, yet analyzing and interpreting these data is a significant challenge. New statistical and computational methods are needed to identify relationships between high-dimensional proteomics and genomics data, imaging data, and information about a person’s environment, diseases, clinical interventions, and health outcomes. Understanding cancer and disease heterogeneity will require new computational methods for third-generation genetic sequencing and related single-molecule technologies. The panel will discuss novel statistical methods in data-mining for cancer and health genomics, the utility of mobile apps in asthma care and research, and single-cell RNA-sequencing in ovarian cancer biomarker development.
Humans are altering both marine and terrestrial ecosystems in substantial ways, but the impacts themselves are not well understood. This is due at least in part to the challenge of collecting representative and reliable data. The ocean environment, for example, makes it difficult to observe species interactions and population dynamics directly. Yet a fairly complete understanding of these processes is essential for effective management and policy. Forests and wildfires present another challenge for the science-based decision-maker. Wildfire is a naturally occurring event; but with increased human presence in and near forests, the risk to human life and property is increasing, and effects on the habitats of forest-dwelling species are also changing. Mitigating the effects of increasing wildfire hazards, while recognizing its important ecological role, requires a deep understanding of the fire regime and how it is evolving. Data for modeling are often mismeasured and subject to size-biased sampling effects. In this session, stochastic modeling and inferential techniques for resolving some of these issues are described, and the speakers demonstrate how these methods are being presented to policymakers and translated into science-based policies.