|
|
|
||
|
|
|||
|
Superior Business Decisions Through Better Data Analysis |
|||
|
|
|||
|
|
Over the last three years, Cox Associates has achieved
breakthrough results in projects completed in each of its practice
areas. In addition, we've been awarded a number of patents in areas
related to our practice. To view summaries of our activities or to read
about our patents, click on the link below for the appropriate discipline. Customer Data
Mining Services Best
Paper Award, 2011 A paper on computational modeling of Chronic Obstructive Lung Disease
(COPD) won a 2011 Best Paper Award from the Society for Risk Analysis (www.sra.org/journal_best_paper_awards.php).
The paper describes the onset and progression of this lung disease in terms
of a cascade of activations of physiological feedback loops. The
paper is available from: www.ncbi.nlm.nih.gov/pubmed/20846171. Improving
Terrorism Risk Analysis A collection of papers on Advances in Terrorism Risk Analysis,
co-edited by Tony Cox and Michael Greenberg, has been made available for
free by the publisher, John Wiley & Sons, at: http://onlinelibrary.wiley.com/journal/10.1111/%28ISSN%291539-6924/homepage/custom_copy.htm New Book on Risk Analysis
of Complex Systems In 2009, Springer published a new book, summarizing recent
research at Cox Associates on Risk Analysis of Complex and Uncertain
Systems. The book introduces principles of good risk assessment and
effective risk management for social, biological, and engineering systems
that are too complex or too uncertain to model in detail, and demonstrates
the practical application of these methods to important, challenging problems
such as terrorism, telecommunications network
design, import risks, carcinogen risk assessment, and antibiotic resistance. Innocentive Award, 2008 In 2008, Dr. Cox's proprietary solution to a
challenge on “Statistical Methods to Predict Clinical Response”
won an InnoCentive Award from Eli Lilly (http://www.innocentive.com/servlets/project/ProjectInfo.po?s=AW). The solution applied data mining and
modeling methods originally developed by Cox Associates for predicting the
future behaviors of telecommunications customers, to predict more quickly and
accurately the clinical responses of patients to therapeutic treatments based
on multiple markers of responses at early time points. Outstanding Risk Practitioner Award, 2007 In 2007, the Society of Risk Analysis (SRA) (http://www.sra.org/) awarded its
"Outstanding Risk Practitioner Award" to Dr. Tony Cox of Cox
Associates, in recognition of "excellent performance in the Practice of
Risk Analysis". Upon being
presented with the award at a ceremony at the SRA Annual Meeting, Dr. Cox
warmly thanked the SRA, but noted that the field contains many remaining
unsolved conceptual problems, and enormous untapped opportunities for
additional practical applications to improving risk management of complex
social, business, and engineering systems. Best Published Paper of 2006 Applying Risk Assessment
In 2007, the Risk Assessment Specialty Section of
the Society of Toxicology (SOT), presented its award for "The Outstanding Published Paper in 2006
Demonstrating an Application of Risk Assessment" to Dr.
Cox for the paper "Estimating Preventable Fractions of
Disease Caused by a Specified Biological Mechanism: PAHs
in Smoking Lung Cancers as an Example" (Cox and Sanders, 2006, http://www.blackwell-synergy.com/doi/abs/10.1111/j.1539-6924.2006.00785.x).
Dr. Cox and his co-author, Dr. Edward Sanders of Philip Morris International
(PMI) in Cox Associates inducted into
In May, 2006, Cox Associates was
greatly honored to be inducted into the
New Book On Quantitative Health Risk Assessment Methods
In November, 2005, Dr. Tony Cox’s new book Quantitative Health Risk Analysis Methods: Modeling the Human Health Impacts of Antibiotics Used in Food Animals was published by Springer. This book provides practical methods for quantifying and comparing uncertain health risks using available information, despite real-world limitations and gaps in scientific knowledge and relevant exposure and dose-response data. The methods are developed and illustrated in detail for the often-controversial field of antimicrobial risk assessment. For more on the book, see: http://www.springer.com/sgw/cda/frontpage/0,11855,4-40517-22-50492725-0,00.html.
How To Improve Qualitative Assessment?
Many qualitative risk rating systems have been proposed that use ordered labels such as “High”, “Medium”, and “Low” to rate the components of risks (e.g., release potential, exposure potential, response probability, and consequence severity) and that then assign an overall qualitative rating to risk based on these component ratings. In July, 2004, applied mathematics research at Cox Associates identified some of the inherent limitations of any such system, proving that, under natural conditions, qualitative risk assessment necessarily makes rating reversal errors (by assigning higher qualitative labels to quantitatively smaller risks) and can produce high qualitative risk ratings even for very small risks. The value of information (VOI) from a qualitative risk assessment can be zero in many practical cases, because the intrinsic high error rate of such systems makes their results untrustworthy. However, augmenting or replacing purely qualitative ratings with simple, robust quantitative variables and rating functions usually resolves these problems, even when the inputs are only imprecisely known.
How Should Combat Plans Be Updated as Situations Change? In 2004, Cox Associates helped Systems View (www.systemsview.com) create and deliver to the US Air Force a new set of dynamic hierarchical planning and decision algorithms for adaptively allocating aircraft to bases and missions – and then intelligently re-allocating them as new information becomes available. The new adaptive planning algorithms help combat planners hedge their bets against risks while exploiting newly perceived opportunities and defending against new threats from intelligent adversaries who are also dynamically re-planning. These planning algorithms, based on a combination of artificial intelligence techniques and mathematical optimization heuristics, have potential applications in many settings where limited resources must be allocated among multiple regions over time while both the current situation in different locations and intelligence about them are evolving.
A Simpler Approach To Dynamic Risk Management
In April, 2004, Cox Associates researchers completed an internal R&D effort to develop a computationally practical, radically simplified approach to making effective current planning decisions – e.g., about inventory, capacity additions, or infrastructure investments – when the future is highly uncertain, e.g., due to uncertain future demands, technology options, costs, and competition. Development and computational implementation of the new approach, led by Dr. Djangir Babayev of Cox Associates, is based on the insight that using a detailed simulation model to generate random samples of future scenarios and then solving for the best current action for each scenario using commercial-quality deterministic solvers can, when combined with risk analysis methods, identify the best current action to take while hedging against realistic uncertainties about which future scenarios will turn out to be true. Unlike earlier scenario-based and stochastic optimization approaches, the new one does not require prior identification of the scenarios to be considered (they are generated via simulation and may be hard to envision or anticipate intuitively), has nice mathematical properties (guaranteeing that approximately optimal current actions and contingency plans will be discovered with high probability), and is computationally much less demanding (since it exploits sampling to overcome the need for explicit combinatorial optimization.) The new approach has the potential to allow existing deterministic solvers for complex planning problems to be re-used to solve new versions of the problems in which future uncertainties are taken into account. This capability will be featured in future planning software tools from Cox Associates to allow our customers to optimize current decisions while hedging against future risks. Society
For Risk Analysis Best Paper Award, 2003 In December of Nonlinear
Breakdown of Hematopoiesis In September of 2003, Cox
Associates delivered to the American Chemistry Council a new biomathematical simulation model of the effects of
mixtures of volatile organic compounds (including benezene)
on hematotoxicity and cancer risk. This model
integrates an MVK-type stochastic two-stage model of carcinogenesis with a
detailed pharmacodynamic model of hematopoietic
stem cell kinetics in the presence of hematotoxic
cell-killing. A dramatic and unexpected finding is that the simulated
hematopoietic system exhibits catastrophic failure modes, leading to an
abrupt transition from relatively low cancer risks to relatively high ones,
as any of three exposure parameters – dose rate (ppm-hours
per day of exposure) or days per week of exposure or number of weeks of
exposure – passes a threshold (whose value depends on the remaining two
exposure parameters). Such bifurcation of risk as a function of
exposure parameters has potentially profound consequences for how exposure-reponse data should be summarized and analyzed. It
implies that traditional dose-response relations (e.g., with ppm-years on the x-axis and probability of cancer on the
y-axis) contain inherent ambiguities, as the aggregate exposure can
correspond to risks that differ by an order of magnitude depending on the
detailed time pattern of dosing. (http://www.birenheide.com/sra/2003AM/program/singlesession.php3?sessid=WRT2&order=3#3
) Risk
Assessment And The Future Of Animal Antibiotics In
September of 2003, Tony Cox presented an invited talk on the future of
antibiotic use at the Seventh Discover Conference on Food Animal Agriculture
(http://www.adsa.org/discover/7thDISCOVERProg4-8-03_files/INTERPRETIVE%20SUMMARIES/Cox%20abstract.htm).
This talk introduced the Rapid Risk Rating Technique (RRRT) and applied it to
show that (a) Human health risks from
continued use of common animal antibiotics such as macrolides
are typically on the order of less than 1 excess case per year; and (b)
Human health benefits from continued use of common animal antibiotics are
typically on the order of more than 1000 cases per year prevented. If
these calculations are even approximately correct, then current regulatory
concerns and efforts to restrict animal antibiotic use may, paradoxically,
cause much more human health harm than they could possibly prevent,
increasing the need to use human antibiotics and hastening the development of
resistance. This highlights the importance of using sound, data-driven,
quantitative risk assessment, rather than qualitative concerns and good
intentions, to inform public policy.
Selected Recent Papers and Presentations Cox LA
Jr. and D Babayev. Networked facilities
expansion problem. International Journal of Information Technology
and Decision-Making 2006, June; 5(2), in press. http://www.worldscinet.com/ijitdm/ijitdm.shtml
Cox LA Jr. Universality
of J-Shaped and U-Shaped dose-response relations as emergent properties of
stochastic transition systems. Nonlinearity in Biology, Toxicology
and Medicine 2006 3(3), in press. Cox LA Jr, Wong C. State transition model for customer relationship
management. Direct Marketing Association Analytics. April,
Bier V, Cox LA Jr. Probabilistic Risk Analysis for
Engineered Systems, in Advances in Decision Analysis. W. Edwards, D.
von Winterfeldt and R. Miles, Eds. Cambridge
University Press, Cox LA Jr. Health Risk Analysis for Risk
Management Decision-Making. Advances in Decision Analysis. W. Edwards, D. von Winterfeldt and R. Miles, Eds. Cambridge University
Press, Cox LA Jr., Popken DA.
Quantifying potential human health impacts of animal antibiotic use: Enrofloxacin and macrolides in chickens. Risk Analysis 2006
26(1), forthcoming. Cox LA Jr. Some limitations of a proposed linear model for antimicrobial risk management. Risk Analysis. 2005 Dec; 25(6): 1327-1332. http://www.blackwell-synergy.com/doi/abs/10.1111/j.1539-6924.2005.00703.x Cox LA, Babayev D.
Optimization under uncertainty via random sampling of scenarios II. Applied
and Computational Mathematics, 2005;4(1): 20-28 Cox LA Jr., VanSickle
JJ, Popken DA, Sahu
R. Optimal tracking and testing of US and Canadian herds for BSE:
A Value-of-Information (VoI) approach. Risk
Analysis, 2005; 25(4): 827-840. http://www.blackwell-synergy.com/doi/abs/10.1111/j.1539-6924.2005.00648.x Cox LA Jr,
Babayev D, Huber W.. Some limitations
of qualitative risk rating systems. Risk Analysis, 2005
Jun;25(3):651-62 http://www.blackwell-synergy.com/doi/abs/10.1111/j.1539-6924.2005.00615.x
Ricci PF, MacDonald TR, Cox LA Jr. Precautionary decision making: Analysis and results. Int. J. Risk Assessment and Management, 2005 6(2-4):237-270. Cox LA Jr. Potential human health benefits of antibiotics used in food animals: A case study of virginiamycin. Environment International, 2005 May;31(4): 549-563. doi:10.1016/j.envint.2004.10.012 Cox LA Jr,
Ricci PF.. Causation
in risk assessment and management: Models, inference, biases, and a microbial
risk-benefit case study. Environ Int. 2005 Apr;31(3):377-97. Cox, LA Jr. Predicting and optimizing customer
behaviors. Chapter Popken, DA and LA Cox. A simulation-optimization approach
to air warfare planning. Journal
of Defense Modeling and Simulation, 1(3), 127-140.
December, 2004., http://www.scs.org/pubs/jdms/vol1number3/Popken.pdf
Cox,
LA Jr. BSE Update. Invited presentation, R-CALF Annual National
Convention. http://www.lmaweb.com/infonewspast.html , http://www.hpj.com/bsetimeline.cfm
Cox,
LA Jr. Dynamics of u-shaped and n-shaped dose-response in a simple model
of cytotoxicity-mediated carcinogenesis.
Poster presented at the Fourth Annual International Conference on Hormesis: Implications for Toxicology, Medicine,
and Risk Assessment.
Cox
LA Jr., Approaches to Antimicrobial Risk Analysis in Food Safety Decision
Making in Poultry Medicine. Invited talk, 2005 AAAP/AVMA Annual Meeting.
Cox
LA Jr., Modeling Lung Carcinogenesis 1: Genotoxic
and epigenetic events in two-stage clonal expansion
(TSCE) models. Invited presentation, Philip Morris Research
Laboratories Symposium on Genotoxic and Epigenetic Tumorigenesis. Crown Plaza Hotel.
Cox LA
Jr., Modeling Lung Carcinogenesis 2: Toward a biologically based model
of smoking and lung cancer . Invited
presentation, Philip Morris Research Laboratories Symposium on Genotoxic and Epigenetic Tomorigenesis.
Crown Plaza Hotel.
Cox
LA Jr., What Fraction of Disease Can be Prevented by
Removing Specific Exposures? Presentation at the Society for Risk Analysis (SRA) Annual
Meeting.
Risk analysis and public health – How to get from good
intentions to good results: Animal antibiotics and other examples. Invited talk, Café Scientifique.
Cox LA Jr., Using
data mining to predict and to influence customer behaviors: Some real-world
examples. Invited
talk, Seminar in Information
Systems and Data Mining,
|
