SAMSI Brings Together Researchers in Optimization for Opening Workshop

20160830_091323Many problems in mathematics, statistics, science, engineering, and everyday life revolve around the choice of a best selection to achieve a specified goal: finding the fastest way to the airport, or the best rice cooker for under $100. From a mathematical point of view, optimization often amounts to finding the maximal value of a function.

This vibrant SAMSI program has produced an unprecedented number of 13 research working groups, which are concerned with fundamental methodology and computational methods for optimization, and applications of optimization to radio therapy; decision analysis; energy and the environment;  and electronic structure models in physics, chemistry and materials science; among many others.

It is the mission of the 2016-17 SAMSI Optimization program to capitalize on and advance this synergy. The program aims to guide the interaction between mathematics and statistics, so as to produce benefits for each area individually, but also combined.

More than 90 students and postdocs attended the Summer School in August 2016, which featured a lively mix of tutorials and hands-on interactive labs, where participants were introduced to state-of-the-art software. The Opening Workshop two weeks later signaled the official start of the research program. The number of participants, more than 130, was limited only by the seating capacity of the lecture room. There was a large variety of presentations, with speakers from academia, of course, but also industry (the oil and gas company ExxonMobil, and the online advertising company MaxPoint Interactive Inc.) and the National Labs (Argonne, Sandia-Livermore, and Sandia-Albuquerque). A special two-hour session gave participants a glimpse at the challenging research problems faced by the National Labs.

Much like attendees from the Summer School, feedback from the participants was overwhelmingly positive.20160829_174219

A number of mid-program events are in the works. The Workshop on the Interface of Statistics and Optimization (WISO) in February 2017 is planned as a high-profile event. It will be lived-streamed for a national and international audience and will feature the pioneers in this area, by giving their work prominent visibility to a broad audience.

Overall, the researchers in attendance got a chance to see how their specialized research could contribute to these industries. The Optimization Program runs from August through the end of May in 2017. For those interested in learning more about this program and its various working groups, visit: www.samsi.info/opt.

SAMSI 2016-2017 ASTRO Year-long Program Begins with Bang

This composite image shows an exoplanet (the red spot on the lower left), orbiting the brown dwarf 2M1207 (centre). 2M1207b is the first exoplanet directly imaged and the first discovered orbiting a brown dwarf. The photo is based on three near-infrared exposures (in the H, K and L wavebands) with the NACO adaptive-optics facility at the 8.2-m VLT Yepun telescope at the European Southern Observatory Paranal Observatory.
This composite image shows an exoplanet (the red spot on the lower left), orbiting the brown dwarf 2M1207 (centre). 2M1207b is the first exoplanet directly imaged and the first discovered orbiting a brown dwarf. The photo is based on three near-infrared exposures (in the H, K and L wavebands) with the NACO adaptive-optics facility at the 8.2-m VLT Yepun telescope at the European Southern Observatory Paranal Observatory.

Since the dawn of time we all have often looked at the night sky and wondered WHAT, if anything, is out there?

In this ongoing 2016-2017 yearlong SAMSI Program on Statistical, Mathematical and Computational Methods for Astronomy (ASTRO), astrophysicists, mathematicians and statisticians are working together among many other things, to explore better ways to find the existence of other planets, in particular the ones which have habitable conditions as our own planet Earth.

This year’s opening workshop for the ASTRO program was held at the NC Biotech Center from August 22-26 which brought together some of the most brilliant minds in the field to discuss among other research topics, the possibility and existence of other worlds or “exoplanets.” The workshop featured a multitude of talks and panel discussions on the various research topics that includes Astrophysical Emulation, Astrophysical Populations (exoplanets), Gravitational Waves, Synoptic Time Domain Surveys and Cosmology.

Over 90 participants from around the nation and also a from other countries (Canada, Spain, UK), specializing in astronomy and astrostatistics were present for the five day workshop that featured speakers from NASA, Caltech, Harvard, Penn State and Yale just to name a few. The year-long program will allow astrophysicists, mathematicians and statisticians to collaborate via virtual media (e.g., weekly webex meetings) and they will be working together for approximately the next nine or so months to analyze huge size data and explore better ways to improve current methodologies based stellar observations  produced by spectrographs and other ground-based and space-based astronomical surveys.

20160823_102415
– The workshop featured numerous speakers who talked about the growing importance of exoplanet research.

Currently, one of the emphases in the astronomy field is to find the existence of exoplanets and/or other worlds that have the potential to support life. As this is a hot topic in the astronomy field, scientists and mathematicians are focusing their efforts on finding these exoplanets right here in our own galaxy. By trying to locate exoplanets, the potential exists for probes to be sent to explore these regions for earth-like planets.

Since 1988, survey based analyses have identified the discovery of more than 3,500 exoplanets. The data provided for these discoveries came from the High Accuracy Radial Velocity Planet Searchers (HARPS), beginning in 2004, and later by the Kepler Space Telescope launched in 2009.

Eric Fiegelson, a Distinguished Senior Scholar and Professor from Penn State’s Astronomy and Astrophysics Department, was one of the attendees and was one of the speakers for the opening workshop. Fiegelson was extremely excited about this opportunity to work with other researchers in order to learn how both astrophysicists and astrostatisticians can bring their collective experience and knowledge to the table in order to potentially lead to the discovery of other exoplanets.

Over the past 25 years Fiegelson has been involved in astronomy and teaching, he said, “This event was the first time I have ever seen a room filled with nearly 50% astronomers and 50% statisticians…SAMSI made this possible!” Fiegelson explained why this was significant because until now, the two disciplines in the science of astronomy rarely worked together on a grand level research endeavor like this. Fiegelson is also one of the many visiting fellows at SAMSI for this program whom are charged with supporting the research and collaboration of this program from this consortium of brilliant minds in the field of astronomy and statistics.

20160826_105800
– Astrostatisticians met and collaborated with astrophysicists and astronomers to work on ways to improve data analysis in exoplanet research.

It was only fitting that while this workshop was in session, astronomers announced that they may have found a planet 1.3 times more massive than Earth. The exoplanet is known as Proxima B and current analysis suggests that it resides near the star Proxima Centauri, our sun’s nearest neighbor. Proxima B is within the habitable zone to Proxima Centauri, which means that the exoplanet can support liquid water given sufficient atmospheric pressure and therefore has the potential to sustain life. Proxima B is approximately 4.7 million miles away and would take almost 20 years to reach with our current technology of space exploration. Still the existence of Proxima B is our most hopeful prospect yet of finding other life out there in the cosmos.

The news of this exciting discovery was well received by those attending the opening workshop. The existence of the very source of their research further supplanted the need to explore this topic even more. Many of the researchers are excited for the chance to work together and learn each other’s capabilities. Overall, the hope for the ASTRO program is that it provides a wealth of opportunities by promoting the sharing of data and ideas and by allowing scientists to collaborate almost on a daily basis for nine months that could potentially have huge ramifications into the research of five focused research topics.

The ASTRO program, which started in August of this year will be ongoing through May, 2017. To see what other interesting topics and workshops will be discussed in this program, visit: www.samsi.info/astro.

20160826_110659
– Over 90 participants from around the nation and also a from other countries (Canada, Spain, UK), specializing in astronomy and astrostatistics were present for the five day workshop that featured speakers from NASA, Caltech, Harvard, Penn State and Yale just to name a few. The researchers met from all over to explore better ways to find the existence of other planets, in particular the ones which have habitable conditions as our own planet Earth.

SAMSI Deputy Director to Deliver Helen Barton Lecture Series at UNC-G

Dr. Sujit Ghosh, Deputy Director of the Statistical and Applied Mathematical Sciences Institute (SAMSI), has been invited by the University of North Carolina-Greensboro’s Department of Mathematics and Statistics to present a series of three lectures this fall as part of the Helen Barton Lecture Series in Mathematical Sciences.

The lecture series has been a fixture in the academic community since 2012 and the target audience for these talks are graduate and upper level undergraduate students and faculty members. Dr. Ghosh is one of many distinguished mathematicians/statisticians who have been invited to speak for the series.

Ghosh’s three-part series, entitled, “Statistical Inference Subject to Shape Constraint,” will take place on the UNC-G campus from, Monday, November 14 thru Wednesday, November 16.

The focus of Dr. Ghosh’s talk will be to present an introductory overview of lectures on statistical inference for density and regression function estimations that are known to preserve a set of shape constraints. Some popular applications include the study of:

  • utility functions, cost functions, and profit functions in economics
  • the analysis of growth rates as a function of various environmental factors
  • the study of dose response curve in the phase I clinical trials
  • the estimation of the monotone hazard rates and the mean residual life functions in reliability and survival analysis and many more

In addition to theoretical results and applications, the lectures will also feature demos of R software packages that can be used to compute various statistical data and graphics.

Ghosh has served as the Deputy Director at SAMSI since 2014. He has served as the Co-Director of Graduate Programs in Statistics at North Carolina State University, where he managed over 150 students annually from 2010 – 2013. Before serving in his current role at SAMSI, Ghosh served as the Program Director in the Division of Mathematical Sciences within the Directorate of Mathematical and Physical Sciences at the National Science Foundation from 2013 – 2014.

Prof. Ghosh has more than 20 years’ experience in conducting, researching and applying statistical analysis of biomedical and environmental information in a wide variety of capacities and subjects. On top of these accomplishments professionally, he has a lengthy and extensive academic record which includes: giving over 125 invited lectures at seminars and national/international meetings; serving as a statistical investigator and consultant for over 40 different research projects funded by numerous private industry leaders and federal agencies and publishing over 95 referred journal articles in the area of biomedical, econometrics and environmental sciences just to name a few. Dr. Ghosh has also co-edited a popular book entitled “Generalized Linear Models: A Bayesian Perspective.”

To see more information on Dr. Ghosh’s lecture or other upcoming events visit the web page for the Helen Barton Lecture series.

IMSM 2016 Prepares Graduate Students for ‘Real World’ Research

The sun set on a hot July day across the street from North Carolina State University, signaling the end of another positive experience in research.

Nearly 40 Graduate Students, of various science, applied mathematics backgrounds and statistics celebrated their accomplishments and experiences after attending the 2016 Industrial Modeling Workshop (IMSM) for Graduate Students in Raleigh, N.C., July 18-27.

This year marked the 22nd anniversary of the IMSM workshop, a major educational outreach component of the Statistical and Applied Mathematical Sciences Institute (SAMSI). Each year, SAMSI invites graduate students from across the country to attend a 10-day workshop, where various industrial and government agencies partner with academia to solve “real world” problems that impact our lives.

This year, SAMSI was pleased to have representatives from: Sandia National Laboratories; Rho, Inc.; the US Army Corps of Engineers (USACE); Environmental Protection Agency (EPA), Pfizer and the Cooperative Institute for Climate and Satellites (CICS). The IMSM workshop is sponsored by SAMSI as well as the Department of Mathematics and the Center for Research in Scientific Computation (CRSC) at N.C. State University.

Graduate students were split into six teams and presented with six different projects from the various industry and lab partners. Subjects of these problems ranged from climate and health to environmental issues. Each team was guided by at least one Industry and one faculty mentor who offered support and helpful hints to make sure the team could develop workable solutions within the allotted time frame.

The IMSM workshop introduces graduate students to the effective application of academic knowledge towards solving “real world” problems. Students also learned valuable skills about time management and team-based research in a time-constrained environment – a practice that is key to achieving results in industry and government labs. The group of students was dynamic, representing such disciplines as Geophysics, Engineering, Biology and of course Applied Mathematics and Statistics. The diversity of students played a pivotal role in helping the teams to develop synergy through their collective strengths and experience in order to reach a common goal. Most students were excited about the opportunity to attend and collectively looked forward to the challenges presented in the IMSM workshop. In the end, industry and lab partners as well as the students benefitted from the experience of producing research results that have the potential to advance “real world” applications.

Discussion

One highlight of this year’s projects was a problem set directed at ways to identify elements of various allergens in order to develop therapies against food allergies. This important issue was posed by Rho, Inc.

Based on research from the Centers from Disease Control (CDC), food allergies are specifically prevalent in children ages 5 and above. This trend has increased by 18% from 1997 to 2007 and effects nearly 5% of adults and 8% of children. Primarily, eight foods account for 90% of all food allergy reactions: milk, eggs, peanuts, tree nuts, wheat, soy, fish and shellfish.

The students’ focus was to look at nut allergies. Nut allergies make up more than 25% of the most common foods associated with severe allergic reactions. In this specific case, the research developed here could easily be replicated towards the study of other food allergies as well. Allergies are caused by a person’s immune system overreacting to harmless proteins in our food or the environment.  One tool for analyzing these proteins is a peptide microarray. These microarrays help to identify parts of certain proteins that trigger allergic reactions. Fragments of allergy-triggering proteins are arranged on small plates or “chips” and exposed to a patient’s blood.  Antibodies from the patient’s immune system found in the blood will react with some of the fragments. These interactions can be detected by microscopes or scanning machines. The data from these experiments, however tend to be “noisy” when researchers try to accurately determine which protein fragments react with the patient’s antibodies. The students’ aim was to try to identify a more effective way to clear up the noise in these samples. Clearing up the noise ensures better predictability by the researchers in their analysis.

Nut Allergies

The data from samples presented by Rho, Inc., had positive markers for a specific nut allergen. The students analyzed these samples and created an algorithm that could identify these patterns more quickly. The students identified the outliers in each sample, which correlated into clearing up the noisy data from these findings. Correctly identifying these outliers made the predictions about this data more reliable and accurate. The result of applying this approach led to identifying 96% of the noise or “bad spots” on a microarray. By identifying these bad spots with a high degree of certainty, one can have a more effective tool to correctly see what protein fragments are triggering allergies.

Though this algorithm was a big break through, still much research needs to be done. The students’ assistance was a positive step forward on this problem.  With these new findings, Rho, Inc., can now go back and apply some of these same techniques to their ongoing research for this problem. It is work like this that further justifies the purpose of bringing great minds together in order to tackle some of life’s puzzles and help us all to live more problem free.

USACE presented two problems: one on habitat quality assessments in the Columbia River and the second on using surface wave properties to predict nearshore bathymetry. Bathymetry is a measurement of submarine topography and can be used to indicate changes in the ocean floor. This near shore analysis could prove vital for predicting damage to coastal environments due to major storms or significant erosion. Storm surge and erosion also negatively impact transportation routes and civil infrastructure. Collectively, these factors would prohibit efforts of support agencies to assist the civilian populace with critical needs in an emergency.

The group used USACE data from Duck, N.C., compiled from various resources to determine coastal depths within 500 m of the coastline. This distance is crucial when it comes to large vessels providing logistical aid support. Support agencies want to ensure adequate water depth, keeping these large vessels from running aground in poor conditions.  The data could also help to understand the various impacts of erosion on coastal structures and transportation routes.  Studies like this have been used in other situations as well, like saving the Historic Lighthouse out at Cape Hatteras.

Accurate measurements of bathymetry in nearshore regions using conventional means are difficult to obtain.  Direct measurements are costly and sparse, and the underlying topography is constantly changing.

Currently, obtaining accurate data related to this research requires many man hours and often costly equipment. The students used USACE data on wave height, wave number and ocean depth to understand how information on the wave mechanics can be used to generate a map of the underlying bathymetry. They used mathematical representations of the connections between measurable wave properties and bathymetry to develop a statistical algorithm for estimating the water depths along a one-dimensional profile.

Bathymetry

The students used data provided by remote sensing platforms, compiled from airborne, satellite and onshore sensors. They studied the dispersion relationship connecting water depth to surface properties, including wave length and period, and discovered using these factors as input provided a relatively accurate estimate of the bathymetry.

Using three different inversion methods, the students accurately determined ocean floor topography up to 900m away from shore. The students found that by using these multiple measurement types, it helped to reduce the amount “noise” in a given variable. In addition, the students determined which inversion method was the best algorithm to use when attempting to accurately identify this data.

Though the group was successful in finding a solution, more work is still needed. The researchers suggested more refinement of their selected inversion method in order to account for more parameters such as beach profile and more access to wave number profiles throughout a given year. These factors could help to isolate trends in the shifting of the ocean floor, which could lead to making mitigation efforts to correct these issues easier.

The group’s final recommendation was to apply this information to a higher fidelity model in order to assess bathymetry in multiple dimensions. The USACE industry mentor looked upon the results favorably. The students’ findings have the potential for numerous applications in keeping with the USACE mission at home and abroad.

Dining Out

 

Overall the consensus of the graduate students was that this workshop was helpful in preparing them for their future contributions in research. The IMSM is a valuable tool for industry as well. Industries actively seek qualified up and coming researchers by being a part of workshops like this and the research gained also has the potential to advance the work in their various research. As the workshop closed, the students spent their last night dining together and reflecting on the experiences they shared over the previous week and a half with peers and faculty and industry mentors in the program.

 

Planning and scheduling by SAMSI has begun for the 2017 IMSM; applications for the workshop next year will be accepted in January. To find out more and apply, interested graduate students should visit the SAMSI website at: www.samsi.info/IMSM17.

SAMSI Poised to Help Hone Gravitational Wave Astronomy, Astronomers’ New Sense

February 24, 2016

(Written by the ASTRO program organizing committee)

LIGO_0– A long time ago in a galaxy far, far away, two large black holes—each with a mass of about 30 suns—reached the end of an aeons-long orbital dance. In the final second of their separate existence, they spiraled toward each other, whirling with a frequency that quickly rose from tens to hundreds of cycles per second. At last they touched, then violently merged in the space of about twenty milliseconds, producing a single black hole that quickly settled down to a bloated, lone existence. Had a video camera been present in the vicinity, it would likely have seen little; black holes are black, after all, regions where gravity is so strong that not even light can escape. Yet during that final merger, the power emitted by this event was larger than all of the power being emitted in light by all of the stars in all of the galaxies in the observable universe. The merger shone, not in electromagnetic waves, but in gravitational waves. The black hole binary’s dance continually sloshed the fabric of space and time in its vicinity, sending out waves carrying news of the invisible event as fluctuations in the spatial separations of objects, and in the flow of time. The waves began as gentle ripples during the long inspiral, steadily climbing in frequency and amplitude; they roiled and crashed during the merger; and finally, they decayed away like the ring of a bell. They followed paths outward from the merger in all directions at the speed of light, diminishing in amplitude but maintaining their shape, an encoding of the story of the merger in the dynamics of spacetime. After a billion-year journey, the waves reached Earth.

This is not the start of a science fiction tale. An international team of over a thousand scientists has observed this merger, the culmination of over four decades of effort sponsored by the National Science Foundation (NSF) and international sources. And NSF’s Statistical and Applied Mathematical Sciences Institute (SAMSI) will soon help astronomers to take the next steps to make the most of this and future gravitational wave discoveries.

ligo20160211d-2to1-web

Image Credit: SXS, the Simulating eXtreme Spacetimes (SXS) project (http://www.black-holes.org) 

Gravitational waves and LIGO

In 1916, Einstein realized that the theory of gravity he had proposed a year before—general relativity, a revolutionary reframing of gravitational interaction, not as the consequence of long-range forces, but rather as a consequence of curvature of spacetime—implied the existence of a new type of radiation, gravitational waves. But the theory revealed space to be incredibly stiff, so resistant to changes in curvature that even violent motions of large masses would produce what seemed to be immeasurably small waves. By the late 1970s, scientists in the U.S. and Europe had converged on a vision for how to make the immeasurable measurable. The Laser Interferometer Gravitational wave Observatory (LIGO) is the realization of this vision.

Discovery!

ligo20160211a-discovery

LIGO’s time series data from the binary black hole merger event GW150914; see the LIGO project page, Gravitational Waves, As Einstein Predicted, for details. Image Credit: Caltech/MIT/LIGO Lab

On September 14, 2015, the waves from that distant merger met LIGO and produced a signal. Months of analysis by many dozens of scientists confirmed its reality, and enabled detailed measurement of the properties of the merging black holes, and of the final hole. The LIGO project announced the discovery to the world on February 11, 2016, dubbing the event GW150914. The discovery marked the confirmation of Einstein’s century-old prediction. But more than that, it marked the opening of a new sense with which astronomers could examine the universe.

SAMSI ASTRO and LIGO

SAMSI is poised to help astronomers hone their new sense. In November 2014, SAMSI sought input from the astronomical community for a year-long program that would gather astronomers, statisticians, and applied mathematicians to address challenging interdisciplinary problems in astronomy. Led by statistician G. Jogesh Babu (Penn. State University), a team of scientists identified a set of timely and compelling research directions, under the overarching and overlapping themes of time-domain astronomy and survey-based astronomy. With renovations to LIGO nearing completion, gravitational wave data analysiswas quickly identified as a focus area, along with exoplanets (which are detected via time series measurements), synoptic surveys(an emerging mode of large-scale automated time-domain observing), and cosmology. In September 2015, scientists gathered at SAMSI to plan the 2016-17 Program on Statistical, Mathematical and Computational Methods for Astronomy (ASTRO). The planning team included LIGO scientists who had only just learned of the candidate detection, and had to keep it secret until confirmed.

Of five working groups planned for the ASTRO program, four will address LIGO data analysis challenges, in concert with related challenges in other areas of time-domain astronomy (a fifth working group will focus on statistical problems in cosmology). One working group will study the potential role of new stochastic process models for analysis of time series data from LIGO and exoplanet surveys, particularly models that abandon the simplifying assumptions of stationarity and Gaussianity underlying most currently-used methods. Another working group will focus on gravitational wave and exoplanet signal detection, and how best to use detected signals for demographic studies (for example, to infer the prevalence and diversity of binary black hole systems, and other sources of LIGO signals). A third working group will address the data-theory interface in the regime of computationally expensive theoretical calculations, where it is impossible to directly compute detailed predictions for every candidate model for the data. Numerical general relativity calculations of binary black hole mergers are a motivating example; similar challenges arise in cosmology.

Finally, a working group on synoptic time-domain surveys will address how to find electromagnetic counterparts to gravitational wave sources. Black hole binary mergers, by their very nature, are essentially invisible electromagnetically. But astronomers expect LIGO to detect other types of events that synoptic surveys could capture electromagnetically, providing opportunities for synergistic multimessenger astronomy. These include such exotic phenomena as merging binary neutron stars, and mergers between black holes and ordinary stars, neutron stars, or white dwarf stars. In addition, gigantic stellar explosions, such as those producing supernovae or gamma-ray bursts, may produce detectable gravitational waves. In a tantalizing twist of fate, astronomers have observed all of these types of objects, and presumed that the first LIGO events would come from such already-known systems. Instead, the first LIGO signal was from a type of system hitherto undetected. What other surprises might this new ear on the sky reveal to us?

SAMSI and Astronomy

The ASTRO program is just the latest of several productive programs SAMSI has hosted to build interdisciplinary partnerships between astronomers, statisticians, and mathematicians. The first such program was the 2006 Spring Program on Astrostatistics (also led by Babu). It, too, included working groups addressing problems in gravitational wave and exoplanet astronomy. Many participants built long-lived collaborations at SAMSI; several are helping to organize the forthcoming ASTRO program. SAMSI’s 2012-13 Program on Statistical and Computational Methodology for Massive Datasets included a week-long Workshop on Astrostatistics, organized by Babu, exploring the intersection of astronomy and “big data.” In the summer of 2013, exoplanet astronomer Eric Ford (Penn State University) led a three-week program, Modern Statistical and Computational Methods for Analysis of Kepler Data. It spawned an independent ExoStats2014 workshop, and one of that program’s working groups continues to meet two and a half years later. Finally, the ASTRO program’s working group on inference with computationally expensive models will build on expertise gained from the 2006-07 Program on Development, Assessment and Utilization of Complex Computer Models, and the 2011-12 Programs on Uncertainty Quantification; participants from both of those programs are on the ASTRO planning team.

More information can be found at: www.samsi.info/astro. Contact directorate liaison: Sujit Ghosh at ghosh@samsi.info.

Ghosh Receives Honorary Degree from Thammasat University

February 2, 2016

TU-HD-diploma-engSAMSI’s Deputy Director and Professor of Statistics at North Carolina State University (NCSU), Sujit Ghosh, received an honorary doctoral degree in statistics from Thammasat University (TU) in Thailand.

This is one of the highest forms of recognition a university can offer. Thammasat University primarily gives honorary doctorates to people from Thailand.

Ghosh has been visiting the Department of Mathematics and Statistics at TU since the summer of 2005. “I have offered several short courses (e.g., Bayesian methods, Monte Carlo Statistics, Spatial Statistics, etc.) which have now been incorporated into their doctoral curriculum,” said Ghosh.

In addition to graduate students, the courses were attended by the local faculty from TU and now their faculty are trained to offer such courses on their own.

Ghosh also co-supervised at least 4 doctoral students from TU who initially attended his courses and then worked with him on completing their doctoral dissertations. Three of them visited him at NCSU during the last six months of their doctoral programs to complete their theses. All of them are currently serving as lecturers at renowned universities in Thailand.

“I am truly honored to receive this recognition from Thammasat University. I hope to continue our wonderful relationship,” said Ghosh.

The graduation ceremony took place on November 16, 2015.

Emergency Department Simulator Uses Analytics to Help Administrators Make Data-Driven Decisions

July 25, 2014

flowmapEmergency departments (EDs) are under growing pressure; while the number of ED visits have sharply increased, the number of EDs serving this need has actually decreased. According to a report from Rand Corporation, ED doctors are increasingly becoming the decision-makers regarding hospital admissions. Today, nearly half of all non-obstetrical hospital admissions occur through the ED. With the adoption of the Affordable Care Act, it is expected the number of ED visits will continue to rise. ED staffs are, therefore, looking for ways to make effective decisions to make their departments more efficient.

A group of researchers from the University of Florida and the Statistical and Applied Mathematical Sciences Institute (SAMSI) have created an online simulator to help hospital ED administrators understand how analytics and simulation can be used to inform decisions in the ED. In particular, the simulator reveals how various factors or decisions affect the flow of patients through the ED. The group includes, Kenneth Lopiano, SAMSI; Joshua Hurwitz, Jo Ann Lee, Scott McKinley, James Keesling, University of Florida Department of Mathematics; and Joseph Tyndall, University of Florida Department of Emergency Medicine.

The simulator is freely available on the web at http://spark.rstudio.com/klopiano/EDsimulation/. On the website doctors or administrators can change several different variables to best mimic the conditions in their particular ED. For example, one can change the number of beds, number of doctors, number of nurses for various hours of the day, or number of patients entering the ED at different times of the day.

Lopiano, who was a postdoctoral fellow at SAMSI during this past year’s Data-Driven Decisions in Healthcare research program, learned about the power of simulation in healthcare through SAMSI-sponsored working groups. It was during a visit to his alma mater, the University of Florida, to discuss his SAMSI experiences when Lopiano learned of lead author Joshua Hurwitz’s efforts. There Lopiano connected with former SAMSI postdoctoral fellow and assistant professor Scott McKinley who introduced Lopiano to Hurwitz. Realizing their common research interests, the core research group was formed which led ultimately to the online simulator, principally developed by Lopiano and Hurwitz. The online simulator has seen substantial increases in traffic since the publication of their research paper in BMC Medical Informatics and Decision Making.
The simulator recognizes that the causes of ED crowding are variable and require site-specific solutions. For example, in a nationally average ED, provider availability can cause bottlenecks in patient flow while investments in other resources may not have the positive impact an administrator would expect. Further, the simulator recognizes that by reallocating resources and creating alternate care pathways, some EDs can dramatically expedite care for lower acuity patients without delaying care for higher acuity patients.

Lopiano, co-founder and principal collaborator of Roundtable Analytics, a healthcare analytics company based in Raleigh, North Carolina, said, “A simulator is very effective because it is risky for health systems to implement overhauls in their care-delivery systems. By using a simulator, administrators are able to evaluate many different scenarios without making these costly and time-consuming changes. Most importantly, administrators can understand the consequences of operational decisions, both intended and unintended.”

The paper published in BMC Medical Informatics and Decision Making is available at: http://www.biomedcentral.com/1472-6947/14/50. Kenneth Lopiano may be contacted at klopiano@roundtableanalytics.com.

About SAMSI

The Statistical and Applied Mathematical Sciences Institute (SAMSI) is one of eight mathematical institutes funded by the NSF’s Division of Mathematical Sciences, but is the only one that focuses on statistics and applied mathematics. Its mission is to forge a new synthesis of the statistical and applied mathematical sciences with disciplinary sciences to confront important data- and model-driven scientific challenges. It is based in Research Triangle Park, North Carolina. SAMSI was founded in 2002.

SAMSI is a partnership of the National Science Foundation with a consortium of Duke University, North Carolina State University, the University of North Carolina at Chapel Hill, and the National Institute of Statistical Sciences. You can find more information at www.samsi.info, @NISSSAMSI.

SAMSI Appoints New Directorate Members

June 2, 2014

The Statistical and Applied Mathematical Sciences Institute (SAMSI) is pleased to announce the appointments of three new members of the Directorate.

Sujit Ghosh, Professor of Statistics at NC State University (NCSU) and currently a Program Director in the NSF Division of Mathematical Sciences, will become Deputy Director of SAMSI beginning September 8, 2014. Sujit’s research interests are in area of Bayesian statistical methods for analyzing biomedical, econometrics and environmental models. Ghosh previously participated in several SAMSI programs, including as Faculty Fellow representing NCSU in the 2011/12 program on Uncertainty Quantification. Ghosh received his Ph.D. in statistics from the University of Connecticut in 1996is actively involved in teaching, supervising and mentoring graduate students at the doctoral and master levels. He has supervised over 30 doctoral graduate students and 3 post-doctoral fellows and he has also served as a statistical investigator and consultant for over 40 different research projects funded by various leading private industries and federal agencies. In addition to his time at NCSU, he has been a visiting professor at Thammasat University in Thailand, Bocconi University in Italy, Middle East Technical University in Turkey, Techincal University of Crete in Greece and National University in Singapore. He is an elected fellow of the American Statistical Association and the recipient of the 2008 IISA Young Investigator Award. He has also been elected as the President of NC Chapter of ASA in 2013 and served as the Co-Director of Graduate Programs in Statistics at NCSU managing over 150 students annually during 2010-2013, and the Project Director of a training program for undergraduates funded by the NSF during 2007-2013.

“Sujit brings to SAMSI a mature understanding of SAMSI’s research mission, as well as administrative and grant management experience which will be invaluable as we plan for our next funding cycle,” noted Richard Smith, Director of SAMSI.

Thomas Witelski, Professor of Mathematics at Duke University, specializing in nonlinear partial differential equations and fluid dynamics, will become Associate Director of SAMSI for a three year term beginning July 1, 2014. His expertise will be valuable on the applied mathematics side of SAMSI’s activities, and he will also act as SAMSI’s liaison with Duke University during this period. Witelski received his Ph.D. in Applied Mathematics from California Institute of Technology in 1995. Before working at Duke, he was an NSF Postdoctoral Fellow and an Applied Mathematics Instructor at the Massachusetts Institute of Technology (MIT). He is a member of the Society of Industrial and Applied Mathematics, the American Mathematical Society and Tau Beta Pi. He is also the co-Editor-in-Chief of the Journal of Engineering Mathematics and a Division Editor of the Journal of Mathematical Analysis and Applications. He also serves on the editorial board for the European Journal of Applied Mathematics, Discrete and Continuous Dynamical Series B.

Ghosh and Witelski will replace Snehalata Huzurbazar from the University of Wyoming, whose term as Deputy Director ends June 30, 2014, and Ezra Miller from Duke University, whose term as Associate Director also ends June 30, 2014. To fill the gap between Snehalata and Sujit, SAMSI is delighted to welcome back Pierre Gremaud, Professor of Mathematics at NCSU, as Interim Deputy Director for July and August, 2014. During this period, Pierre will be primarily responsible for the education and outreach side of SAMSI’s activities. Pierre previously served as Associate Director of SAMSI from July 2008 through December 2009, and as Deputy Director from January 2009 through June 2012.

About SAMSI

The Statistical and Applied Mathematical Sciences Institute (SAMSI) is one of eight mathematical institutes funded by the NSF’s Division of Mathematical Sciences, but is the only one that focuses on statistics and applied mathematics. Its mission is to forge a new synthesis of the statistical and applied mathematical sciences with disciplinary sciences to confront important data- and model-driven scientific challenges. It is based in Research Triangle Park, North Carolina. Samsi was founded in 2002. SAMSI is a partnership of the National Science Foundation with a consortium of Duke University, North Carolina State University, the University of North Carolina at Chapel Hill, and the National Institute of Statistical Sciences. You can find more information at www.samsi.info, @NISSSAMSI.

Researchers Help Boston Marathon Organizers Plan for 2014 Race

April 14, 2014

After experiencing a tragic and truncated end to the 2013 Boston Marathon, race organizers were faced not only with grief but with hundreds of administrative decisions, including plans for the 2014 race – an event beloved by Bostonians and people around the world.

One of the issues they faced was what to do about the nearly 6,000 runners who were unable to complete the 2013 race. The Boston Athletic Association, the event’s organizers, quickly pledged to provide official finish times for these runners. Thinking ahead, they also had to consider how to provide these runners with an opportunity to qualify for the 2014 race.

To seek advice on these issues, they contacted Richard Smith, a statistician and marathon runner at the University of North Carolina at Chapel Hill, and director of the Statistical and Applied Mathematics Sciences Institute (SAMSI) based in Research Triangle Park, N.C. They asked Smith to come up with a statistical procedure for predicting each runner’s likely finish time based on their pace up to the last checkpoint before they had to stop.
“Once I got their email,” said Smith, “of course I knew I had to help them.” Smith already knew the organizers, as a result of a previous occasion when he provided advice related to the event’s qualifying times.

Smith quickly assembled a team of fellow analysts that included Francesca Dominici and Giovanni Parmigiani at Harvard School of Public Health, and Dorit Hammerling, postdoctoral fellow at SAMSI, who were in the 2013 race and finished uninjured. The team also included Matthew Cefalu, Harvard School of Public Health; Jessi Cisewski, Carnegie Mellon University and Charles Paulson, Puffinware LLC.

The results, and the method the researchers developed, were published in the April 11 edition of PLOS ONE.

With the help of the Boston Athletic Association, the researchers created a dataset consisting of all the runners in the 2013 race who reached the halfway point but failed to finish, and all the runners from the 2010 and 2011 Boston marathons. The data consist of “split times” from each of the 5 km sections of the course (from the start up to 40 km), and the final 2.2 km. The research team was tasked to predict the missing split times for the runners who failed to finish in 2013.

The researchers adapted techniques used in such contexts as computing missing data in DNA microarray experiments and estimating ratings which Netflix subscribers would have given to movies they had not seen. They proposed five prediction methods and created a validation dataset to measure the runners’ performance by mean squared error and other measures. Of the five, the method that worked best used local regression based on a K-nearest-neighbors algorithm (KNN method), though several other methods produced results of similar quality.

The KNN method looks at each of the runners who did not complete the race (DNF) and finds a set of comparison runners who finished the race in 2010 and 2011 whose split times were similar to the DNF runner up to the point where he or she left the race. These runners are called “nearest neighbors.”

“We had to come up with a method to compare the runners based on the split points up to a certain point of the race and then had to decide how many of the nearest neighbors to examine in order to develop a prediction for the DNF runner that would be based on the different finishing times of these nearest neighbors,” said Smith, who has run the Boston Marathon in the past and will run this year’s race. “We decided to choose 200 nearest neighbors. We also tried 100 and 300 nearest neighbors, but the results changed only slightly and didn’t make them better.”

The Boston Athletic Association decided to grant entry to the 2014 race to anyone who was stopped from completing the 2013 event, so they will have a chance to complete the Boston Marathon after all. But in the course of developing the method, Smith and his colleagues realized there were other uses for the technique.

“We have found that using the KNN method looking at a runner’s intermediate split-time will also be useful in predicting the person’s completion time while the race is in progress,” said Smith. “This can be helpful for relatives and friends to be able to meet the person at the finish line.”

Link to the paper: http://dx.plos.org/10.1371/journal.pone.0093800

From UNC News Services