How can we improve Science, Technology, Engineering, and Math education to encourage careers in Biomedical and Pathology Informatics?

Abstract

The Computer Science, Biology, and Biomedical Informatics (CoSBBI) program was initiated in 2011 to expose the critical role of informatics in biomedicine to talented high school students.[1] By involving them in Science, Technology, Engineering, and Math (STEM) training at the high school level and providing mentorship and research opportunities throughout the formative years of their education, CoSBBI creates a research infrastructure designed to develop young informaticians. Our central premise is that the trajectory necessary to be an expert in the emerging fields of biomedical informatics and pathology informatics requires accelerated learning at an early age.In our 4^th year of CoSBBI as a part of the University of Pittsburgh Cancer Institute (UPCI) Academy (http://www.upci.upmc.edu/summeracademy/), and our 2nd year of CoSBBI as an independent informatics-based academy, we enhanced our classroom curriculum, added hands-on computer science instruction, and expanded research projects to include clinical informatics. We also conducted a qualitative evaluation of the program to identify areas that need improvement in order to achieve our goal of creating a pipeline of exceptionally well-trained applicants for both the disciplines of pathology informatics and biomedical informatics in the era of big data and personalized medicine.

INTRODUCTION

“Big Data,” a popular buzzword dating back to 2001, has rendered a need for savvy information scientists and machine learning experts. IBM states that new skills are needed to fully harness the power of big data, and while courses are being offered to train a new generation of data experts, it will take time to build them into the workforce.[2] In an effort to expedite this process, CoSBBI aims to introduce high school students to the practice of informatics in medicine, including the use of computational techniques to solve biological problems. We believe that CoSBBI, and other similar programs, can be used to select and train this new generation of data scientists, thus preparing them for the workforce with the tools, experience, and professional network required to succeed in the domain.

We reported on our 2013 CoSBBI experience[1] describing the program's mission and curriculum. In the CoSBBI class of 2013, 11 scholars developed and presented projects spanning a broad range of topics including bioinformatics, pathology informatics, computational biology, machine learning, image analysis, pharmacogenomics, and telemedicine. These scholars and their faculty research mentors were encouraged to publish their abstracts in the Journal of Pathology Informatics (JPI).[3] Some of those students are now attending college at the California Institute of Technology, Carnegie Mellon University, University of Notre Dame, University of California at Los Angeles, and our very own University of Pittsburgh. Further, our program graduates continue to demonstrate success in informatics an example of which is an alumnus that won the “Pitt Smash Mash!” student start-up competition for an app that serves as a conduit between students and the University Health services. Another alumnus completed a summer research internship at Duke University and was selected from a competitive pool of applicants to present her research at the American Medical Informatics Association (AMIA) 2014 annual symposium (http://www.amia.org/amia2014/high-school-scholars).

This year, we continued our mission of introducing biomedical informatics through a STEM-oriented research academy. We began with a statement of our goal: To provide young talent with a survey of fundamentals, exposure to current informatics research, and a research internship experience. As we continue to build on these goals every year, we have enhanced the 2014 CoSBBI curriculum to meet the evolving needs of our students; this editorial is a synopsis of these changes, including a qualitative evaluation of the 2014 CoSBBI program.

CLASSROOM INSTRUCTION

As in last year's CoSBBI program, the classroom portion was designed to provide a didactic introduction to biomedical informatics, promote an understanding of research, and expose scholars to career opportunities in the field. The complete syllabus and teaching materials for the 8 weeks CoSBBI program can be accessed at: http://faculty.dbmi.pitt.edu/cosbbi/cosbbi2014/. Once again, we used the online, open-access Translational Bioinformatics compilation (PLOS Computational Biology Collection, www.ploscollections.org/translationalbioinformatics) as the primary textbook. In the first few days, scholars attended orientation sessions and were given guest passes to attend the 2014 National Library of Medicine Informatics Training Conference, which was hosted by the Department of Biomedical Informatics (DBMI) at the University of Pittsburgh. For the remainder of the first 6 weeks, the scholars spent two hours each day in the classroom learning about the fundamentals of informatics in a variety of domains. To better equip students with the necessary skills for completing an independent informatics research project, we implemented a week-long, hands-on programing boot camp. It was conducted during the 1^st week of didactic sessions and was a significant addition to the 2014 curriculum. Following the boot camp, each classroom session was comprised of one instructional hour led by doctoral, postdoctoral, and medical fellows, and 1 h of research presentation and discussion led by academic researchers and industry guests. Lectures in the early weeks covered basics of molecular biology, bioinformatics tools, computational thinking, statistics, and data mining, while lectures in later weeks focused on specific areas of biomedical informatics.

The programing boot camp was designed to provide students with a brief introduction to programing and its various applications within the field of biomedical informatics. More specifically, the course was designed to help students: (1) understand basic programing concepts and how to implement those concepts, (2) recognize some basic programing solutions for real-world applications in biomedical informatics, and (3) explore additional languages and more advanced programing topics with a higher level of comfort. Specific topics covered in the programing boot camp included data types, Boolean logic, loops, data structures, file input/output, functions, and code libraries for bioinformatics. Students were instructed in the Python programing language, which is commonly used in biomedical informatics and other disciplines. The instruction was highly interactive, with short lectures interspersed with demonstrations during the 1^st h, followed by working through programing problem sets in small, assisted groups during the 2^nd h. Problem sets were related to lecture material and covered real-world problems in biomedical informatics (e.g., gene detection, elementary clinical decision support). CodeAcademy (http://www.codecademy.com/) was used to implement the problem sets, as it offers basic Python tutorials, is available online, and requires no installation. In addition, CodeAcademy allowed us to design our own problem sets and create detailed tutorials and error messages to help students work through the assignments successfully.

INTRODUCING SCHOLARS TO RESEARCH

While the classroom sessions were focused on concepts and application, the majority of the scholar's time was reserved for pursuing deeper skills relevant to their individual research project. Upon acceptance into CoSBBI, each scholar was matched to a faculty mentor involved in informatics research at the University of Pittsburgh. The mentor/mentee matches are based on the scholars’ background and interests as stated in their application and the availability of suitable mentors. Every attempt is made to making this matching process synergistic with the scholars stated career goals. We wanted to expose the CoSBBI scholars to ongoing, hands-on scientific research in Biomedical and Pathology Informatics. In addition to the classroom sessions, we wanted to focus on the development of three primary areas of research skills: (1) reading, evaluation, and presentation of current literature; (2) conducting independent research in a timely manner; and (3) communication of research through scientific writing and presentation. Our approach to these three primary areas of development is discussed below.

We felt that it was important to demonstrate the importance of reviewing current literature by teaching the students the necessary skills to read, critically evaluate, and present peer-reviewed papers. Toward this end, scholars selected and presented a peer-reviewed article. In this journal club style session, they were encouraged to select scholarly papers relevant to their specific research question with the help of their faculty mentor.

Midway through the program, weekly meetings began to provide students with a forum to discuss the progress of their individual research and receive feedback from their peers and members of the DBMI. This provided the students with milestones toward completion of their project, along with close mentorship and peer evaluation required to complete a sophisticated research project on time.

Finally, we used several classroom sections to train the students how to communicate their research findings through scientific writing and presentation. Near the end of the program, the scholars applied these skills to write scientific abstracts summarizing their work, which can be found at the end of this editorial. On the final day of the program, the scholars gave oral presentations to an open audience at DBMI and presented posters at the UPCI Academy closing symposium.

PROGRAM EVALUATION

Two summer interns (one high school student [GM] and one college student [RU]) developed a survey instrument [Supplement 1 ^{(51.7KB, pdf)} ] with two doctoral students (KMR, JD-M), and conducted interviews with all 2014 CoSBBI students (n = 9). Interviews were recorded, transcribed, and analyzed thematically to identify major successes and problems with the program. We discuss here important observations that emerged as general themes not only relevant for our own improvement, but also as considerations for any Biomedical or Pathology Informatics Department interested in STEM level outreach.

Supplement 1

Final Computer Science, Biology, and Biomedical Informatics evaluation questionnaire

Overall, students viewed the program favorably, rating CoSBBI an average of 4.1 out of 5. They said they were glad that they had received exposure to the field of biomedical informatics (specifically: career paths, research, and work life as a researcher) while learning key skills such as time management, teamwork, and self-discipline. Engaging lectures involving hands-on demonstrations and activities were considered to be the most interesting. The scholars rated their research projects (3.7 out of 5) and research mentors (4.1 out of 5) favorably, with individual comments of dissatisfaction ranging in the areas of communication, compatibility, choice in project formulation, and feeling overwhelmed or unchallenged. The majority of scholars reported feeling that their project was interesting and engaging, and they were able to see its applicability in the domain of informatics. While the sources of dissatisfaction cited by scholars were quite context-specific and mostly relevant to individual scholar-mentor partnerships, one common remark was that their projects did not pertain to bioinformatics. One student, whose project was in clinical informatics said, “My mentor told me on the 1^st day ‘I don’t do bioinformatics, I have nothing related to that; so that's where the problem lies,’ that my mentor didn’t really deal with bioinformatics as his main study.” This revealed an interesting contradiction we had been unknowingly propagating to the scholars: as presented in our editorial last year[1] we promote undergraduate programs in bioinformatics as a route to a future career in biomedical informatics. However, as most practitioners of biomedical informatics know, bioinformatics is not necessarily the ideal foundation for careers in all areas of biomedical informatics. We delve into this further as we discuss our thoughts toward finessing the pipeline.

Among areas of improvement recommended for our program, one of the recurring themes was a desire for more structure in their daily schedule. The students at other sites at the UPCI summer academy are mostly assigned to laboratory-based projects which involve bench work. In the DBMI, however, projects mostly require solitary work on a computer. We had thought of limiting their classroom time in favor of leaving them more time to focus on their individual research projects; many of our respondents perceived this as “too much free time.” It is interesting to note that many respondents also felt they were not making adequate research progress until late in the program.

Being predominantly involved in graduate level education, we may have been too reliant on a scholar's personal initiative in their individual projects, particularly compared to the demanding schedules that high performing adolescents are used to. One idea to address this issue is to incorporate a schedule of research reporting in the classroom from the very beginning of the program. This will encourage peer mentoring and involvement from early stages of the research project. Another idea is to have mentoring teams for every student. Integrating a multilevel team of postdoctoral fellows and/or advanced graduate students in the faculty mentor's laboratory to be daily comrades and mentors to the scholars will allow closer monitoring in addition to providing a more immersive educational experience.

A common theme in the feedback about classroom instruction also implied an expectation of more structured guidance. In keeping with our efforts to let scholars prioritize research work, we did not have homework or exams on a regular schedule. Problem sets were assigned only for the programing and statistics lectures. Readings were assigned for every topic, but students were not specifically tested on them. From our perspective, this was to be the first step into the world of independence in scholarly pursuit. We purposely allowed scholars to choose areas that they wished to focus more time on, while providing them with exposure at a broader level, as well as direction on where to go for more. Some scholars thrived with this freedom, but many others perceived this as a lack of rigor, wishing for more of a challenge in the classroom. To address this issue, we will consider administering weekly quizzes. It would be important, though, to craft tests that are meaningful toward a cohesive preinformatics foundation. Accordingly, the testing emphasis should be on broader concepts, not pedantry.

FINESSING THE PIPELINE

In service to our priority of contributing to a pipeline for biomedical informatics, we previously[1] highlighted the emergence of undergraduate Bioinformatics Departments. We encourage CoSBBI students to explore bioinformatics as a college major, but we are also aware that there are alternative majors of foundational value to the study of biomedical informatics that may appeal more broadly to students’ career interests. An eloquent opinion piece by Dr. William Hersh discusses this quandary faced in advising students on the most appropriate “preinformatics” college major (http://informaticsprofessor.blogspot.com/search?q=pre-informatics).

Our thoughts on an optimal solution are two-fold: We encourage our students to use their summer at CoSBBI to find what inspires them, and we offer paid summer internships at DBMI for any of them who contact us in following years to pursue further research in biomedical or pathology informatics. Two of last year's students returned as interns, one of whom is majoring in computer science, while the other pursues a premed major with coursework in design and business. Studies have shown that increasing engagement and interest in STEM fields during the precollege years is most effective in cultivating graduates in these fields, even more so than high grades and enrollment in advanced level classes.[4]

Critical to CoSBBI's continued success in encouraging careers in Biomedical and Pathology Informatics is continued mentorship from faculty in both of these disciplines. The DBMI (http://www.dbmi.pitt.edu) and the Division of Pathology Informatics and the Center for Pathology Informatics (http://path.upmc.edu/cpi/) are committed to long-term mentoring of CoSBBI scholars. To date, Dr. Becich has written over 30 letters of recommendation to CoSBBI students and dozens of other letters have been supplied by CoSBBI scholar-mentors. Mentors of CoSBBI scholars have provided ongoing input to college major selection encouraging students to enroll in programs in bioinformatics.[1] In 2013, there were 34 such programs in US colleges and Universities,[1] currently, there are over 60 such programs. Another important area of continued mentorship is the CoSBBI internship program which will be the subject of another JPI article soon to be submitted. The CoSBBI high school scholars are guaranteed paid internships in the DBMI once they successfully complete the 8 weeks summer program. To date, of the nearly 30 students that have participated in CoSBBI, eleven have returned to do paid internships. This internship program will be expanded to include the CoSBBI Innovation Internship (Becich, Becich and Boone, manuscript submitted) which focuses on academic, commercial entrepreneurship. In short, CoSBBI is a unique mentorship program which aims to “pipeline” highly trained students for careers in Bioinformatics, Biomedical, and Pathology Informatics.

CONCLUSIONS

Young scholars exit the CoSBBI program with an exceptional first-hand STEM experience. The experience was enhanced by this year's addition of a computer programing boot camp. The boot camp helped to kick start many of the scholar's research projects and had benefits for both proficient and first-time programers alike. We will continue to fine-tune the curriculum and other aspects of the schedule to best benefit the needs and desires of each scholar. We remain committed to providing the best and brightest high school students the opportunity to find their passion in Biomedical and Pathology Informatics. It is our intention for the program to remain free so that students of all backgrounds may continue to participate. In future years, we hope to have more of our scholars compete in the newly created high school student competition at the AMIA annual symposium and hope for more of them to continue their research projects into their undergraduate years. There is no single solution for inspiring the next generation of scientists, but programs, such as CoSBBI, that introduce high school students to the complex world of STEM are vital. The best evidence of the success of CoSBBI is that over 30% of our students return to do paid internships. The addition of the CoSBBI Innovation Internship will certainly increase the interest in STEM through a focus on commercial entrepreneurship.

Financial Support and Sponsorship

This CoSBBI track of the UPCI Academy is supported by NIH grants R01 LM010950 from the National Library of Medicine, R01 GM100387 from the National Institute of General Medical Sciences, Doris Duke Foundation, the National Cancer Institute CURE Program (3P30CA047904-22S1) and support from the Jack Kent Cook Foundation. In addition, this program would not be possible without the infrastructure and support teams from the DBMI NLM Training Program Grant in Biomedical Informatics (T15 LM007059), the University of Pittsburgh Cancer Institute (UPCI) Cancer Center Support Grant for the Cancer Bioinformatics Service (P30 CA47904), the Clinical and Translational Science Institute Biomedical Informatics Core (UL1 RR024153), as well as from funds and in-kind services from UPCI and DBMI, and the Departments of Surgery, Immunology, Computational and Systems Biology, Gynecology, and Pharmacology.

Conflicts of Interest

There are no conflicts of interest.