Undergraduate Student Research Spotlight

Operations Research and Wildlife Conservation

One of the biggest and often overlooked threats towards the environment is from wildlife poaching. It is an act buoyed by the raging black market trade of animal parts and has devastating consequences affecting wildlife populations and the environment. For instance, the International Union for Conservation of Nature (IUCN) in 2011 declared the Western black rhino extinct and identified poaching as the primary cause. The IUCN red list [1] contains all five remaining rhino species as threatened. The perils of poaching go beyond the basic view that this crime needs to be stopped simply to save the life of an animal. Poaching has a series of environmental, economic, health and social impacts [2] and this results in a need for us to understand that stopping poaching is not just in the interest of animals, but also in the interests of humanity.

A few researchers based out of the United States Air Force Academy have decided to tackle this problem from an Operations Research standpoint. Zachary Blanks, a senior in the department of Operations Research, advised by Captain Solomon Sonya from the department of Computer Science, was selected for the Cadet Summer Research Program (CSRP) during which a small number of cadets get to undertake research at various organizations.

The research stems from work conducted by the Team core and CREATE (Center for Risk and Economic Analysis of Terrorism Events) research group at the University of Southern California (USC) working in conjunction with the Department of Homeland Security to apply game theory and machine learning algorithms to develop more efficient ways of reducing poaching incidents. At its core, their research develops a classification model using machine learning algorithms to aid in forecasting future attacks.  The classification algorithm allows rangers from Uganda’s wildlife authority to better allocate patrolling resources while simultaneously reducing computation time and costs. Specifically, they applied machine learning ensemble methods to enhance the analysis of the predictions.

Ensemble methods are algorithms in machine learning that construct a set of classifiers and then classify new data points by taking a weighted vote of their predictions [3]. Essentially, these methods pool together individual learning algorithms to obtain better predictive performance. Zach proceeded to use these methods to predict the most likely areas of observing poaching with the goal of getting good predictive performance at a lower computational cost. According to Zach, his main contribution was to investigate alternative model options as opposed to the pseudo-logistic regression methods that were currently being employed. Using their predictions from data of non-commercial poaching by month, they were able to generate heat maps to show the most probable locations of observing poaching and as a consequence allow the rangers to better focus their efforts. The researchers chose to not include a picture of the heat map due to the sensitivity of the results impacting poaching activity. The researchers state that their results allow them to predict both by season and by month, but that predicting by month might be more preferable because this allows the rangers to develop more effective patrols to combat poaching given that the target area is smaller and within a more localized timeframe.

Despite the primary limitation of their model, as recognized by the students, is that it is in its current form, it is able to observe where poaching will occur but falls short of predicting where poachers are more likely to attack. For now though, the rangers in Uganda were able to employ Zach’s methods last August to find 12 snares, including one elephant trap, likely saving the lives of quite a few animals.

 

Linear Programming Model for Scheduling Med School Clinical Experiences

Two undergraduate students from the University of Michigan’s Center for Healthcare Engineering and Patient Safety (CHEPS), Anna Munaco and Roshun Sankaran, took it upon themselves to restructure and revamp the curricula at the University Of Michigan Medical School in order to give students a more longitudinal clinical experience. Working with University of Michigan Medical School’s Interprofessional Clinical Experience (ICE) program, Anna and Roshun made tremendous progress in improving the quality of the schedule created and in reducing time and energy spent in creating a feasible schedule.

The University of Michigan Medical School created the ICE program for first year medical students. The school assigns each student to a clinic every semester in their first year. Students shadow a different healthcare professional, within their assigned clinic, twice a month (biweekly) over the course of the semester in order to gain a richer understanding of the healthcare system as a whole and of how each professional’s role contributes to patient care. According to Anna and Roshun, working on this project led to a significant reduction of program administrator workload, as administrators did not have to schedule students by hand anymore. Additionally, medical students were being assigned to clinics more often (as they desired), resulting in meaningful experiences and a better idea of their future specialties and career paths.

While performing these assignments, students were expected to adhere to general rules including ones that required every student to be assigned to exactly one clinic, every clinic to have at least one student assigned to it, only students with cars to be assigned to off-site clinics and students in the Medical Spanish elective to be assigned to an on-site clinic. Subject to these restrictions, the goal of optimally assigning students to clinics lent itself well to a binary integer program. The program was implemented in Open Solver, a free add-in to Microsoft Excel that is capable of solving larger problems than Excel’s built-in Solver.

Apart from the benefits that this research provided in helping the University of Michigan Medical School to more seamlessly integrate its new curriculum and to better educate future physicians, the project also resulted in a long-term collaboration between the University of Michigan Medical School and the Center for Healthcare Engineering and patient Safety and has recently led to additional projects. Moving forward, Anna and Roshun’s primarily plan on refining the model includes creating templates for both the first and second semester models so that new student rosters and clinics can be easily imported and used in anticipation that this further refinement will continue to make the scheduling process easier and more efficient.

The students were advised by William Pozehl (CHEPS Research Area Specialist), Dr. Joseph House (Director of Interprofessional Education), Angela Sullivan (Program Administrator), Dr. Amy Cohn (CHEPS Associate Director).

 

References

[1] http://www.iucnredlist.org\

[2] https://theproblemofpoaching.wordpress.com/2013/08/21/why-is-poaching-such-a-problem/

[3] http://web.engr.oregonstate.edu/~tgd/publications/mcs-ensembles.pdf