Translational Medicine

If biomedical research is a sick patient, is computational modeling the prescription?

Planning for what’s beyond Big Data: Translational Systems Biology as a long-term cure for what ails biomedical research

Print Friendly and PDF
Share story:  

Ebola. Flesh-eating bacteria. Rising health care costs. Scientific fraud. Researcher burnout. Unemployed PhDs. Antibiotic-resistant "killer" bacteria. Failed cancer drug trials. Big Data in Medicine. Google and Facebook getting into health.

Medicine and biomedical science are in the news a lot these days, some news good and promising but unfortunately much of it disturbing. The scope of these issues ranges from fundamental aspects of biology to human nature to societal factors to politics. Are all of these events disparate? Or are there common underlying factors that now manifest in these myriad ways? Is this breadth of issues a reflection of the impact of medicine and health on our lives? And can technology, which represents the bleeding edge of human invention and ingenuity, help bail us out?

A substantial aspect of human intelligence is the quest to find patterns within different phenomena, to be able to generalize, to be able to somehow explain what often appears to be a disorganized and chaotic world around us. At a very general level, this is the purpose of Science (capital "S" intentional). In medicine and health, this desire manifests, as a first clinical step, in the diagnostic process. Much of clinical practice specifically involves the observation of a series of seemingly disparate phenomena, and trying to bring order to those observations through some common, underlying cause. So, perhaps we can apply this approach to the list of health-related headlines above. Can we find something at the root of these issues that we can actually do something about?

How can computer modeling provide a boost to the medical field?

Human intuition is a powerful thing, but the point of Science is to check our guesses systematically and establish enough trust in our intuitive conclusions so that we can treat those conclusions as useful knowledge. Even scientists often lose sight of this important role of science, especially since science and technology have become incredibly complex. Fortunately, modern science has recognized that when dealing with complex problems, our little grey cells (á la Hercule Poirot, known for his analytical capabilities) need some help, and we've increasingly turned to computers to augment our reasoning and analytical capabilities.

Elsevier Connect Contributors

Yoram Vodovotz, PhDDr. Yoram Vodovotz (@yoram66) is a Professor of Surgery, Immunology, Computational and Systems Biology, Bioengineering, Clinical and Translational Science, and Communication Science and Disorders at the University of Pittsburgh School of Medicine. He has been studying inflammation and immunity throughout his academic career. Along with Dr. Gary An, he has developed the concept of Translational Systems Biology, which incorporates simulated clinical trials and patient-specific predictive models, with the ultimate goal of patient-specific, self-adaptive control of inflammation. Dr. Vodovotz is a co-founder and Past President of the Society for Complex Acute Illness, which is dedicated to applying complex systems and other computational approaches in sepsis, trauma, and related diseases. He is also a co-founder of Immunetrics, a Pittsburgh-based company that is commercializing this mathematical modelling work. He has published nearly 250 scientific manuscripts.

Gary An, MDDr. Gary An (@Gary_An) is an Associate Professor of Surgery at the University of Chicago, where he is the Co-Director of the Surgical Intensive Care Unit and a Senior Fellow in the Computation Institute of the University of Chicago.

Dr. An is an early proponent (1999) of agent-based modeling, a computational modeling methodology, for biomedical research, and is President of the Swarm Development Group, one of the first organizations created to support and promote the use of agent-based modeling as a general purpose research tool.

Along with Dr. Yoram Vodovotz, he developed the concept of Translational Systems Biology in the mid-2000s, which incorporates computational dynamic knowledge representation, in silico clinical trials and patient-specific predictive models to accelerate the acquisition and utilization of biomedical knowledge to enhance human health.

Translational Systems BiologyDr. An is a co-founder and current Board Member of the Society for Complex Acute Illness, created for the purpose of developing multi-disciplinary approaches involving complex systems analysis and dynamic computational modeling of diseases involving acute inflammation, such as trauma and sepsis. He has served on the editorial boards of multiple journals, and has authored over 100 peer-reviewed journal articles, scientific reports and book chapters.

Drs. Vodovotz and An are the co-authors of the book Translational Systems Biology: Concepts and Practice for the Future of Biomedical Research, just published by Elsevier.

The representation and evaluation of knowledge using computational models underlies modern engineering and has transformed how we perceive our world of mechanical and physical objects. We propose that computational modeling can do the same thing for how we maintain and repair our bodies, thereby fundamentally transforming the medical field as well. On one hand, this is not a novel thought; many very smart people are finding applications of computing to various aspects of health (e.g., the Big Data movement, with Google and Facebook recently throwing their hats into the healthcare ring). But rather than chipping away piecemeal at the edges of issues involving health and medicine (as is currently being done), can we get to the root cause of these problems? In other words, can we treat the underlying cause of these symptoms rather than just trying to alleviate the symptoms themselves?

To jump to the end of the story (Spoiler Alert!), we meld our medical diagnostic approach with the processes of engineering to suggest that what is critically needed is the ability to build better drugs – and that the wider and smarter application of computational modeling is the means towards that goal. Yes, this sounds simplistic, but bear with us for a moment. The benefit of getting better drugs is painfully obvious for several of the issues above: Ebola, cancer, flesh-eating bacteria. Advancing the scientific process of obtaining better drugs also helps us address another set of issues: obviating antibiotic-resistance, clarifying "alternative medicine," utilizing Big Data effectively.

We would argue that infusing principles of engineering – central among them the nearly ubiquitous role of mechanistic computational modeling – into the current scientific process will also have a major social impact: reducing researcher burnout (by reducing the perceived/real arbitrariness of academic funding) as well as scientific fraud (by changing the incentive structure associated with the current arbitrariness of academia). And the overall social impact would be enhanced by the carryover to economic factors: improvement in the efficiency in the drug development pipeline means more drugs, less cost distribution across failed drugs, more specific drugs with reduced side effects, more drugs for each disease to increase competition, more competition to reduce cost, etc.

Unfortunately, like many disease processes, the underlying problem – our inability to effectively turn what we think we know into effective interventions – only feeds on itself, leading to cascading downstream consequences. So, is hoping for a better way to get better drugs a pipedream? After all, we have whole scientific communities and a large, powerful industry aimed at precisely that goal. But is this structure, in of itself, part of the problem?

There are certainly signs, undercurrents and not-so-undercurrents running through our society that would cause us to think that there are inherent barriers within these large bureaucracies. Books like Think Like a Freak by Steven D. Levitt and Stephen J. Dubner and Zero to One: Notes on Startups, or How to Build the Future by Peter Thiel and Blake Masters (and there are many others; apologies to those authors) have pointed out the ways by which historically/culturally embedded thought patterns can limit our ability to solve problems. A recent article in The Atlantic, "Why Experts Reject Creativity," neatly summarizes our paradoxical relationship with innovation and creativity.

We contend that the biomedical research community is subject to these same forces. We further posit that mechanistic computational modeling is the tool by which truly innovative science can be separated from politically-expedient/trendy science.

So, here follows what can be thought of as an "origin story" – tracing how one potential way of addressing the myriad challenges of biomedical science came about: the blending and leveraging of technology, specifically computer modeling and simulation, to address bottlenecks in the scientific process that inhibit our ability to put our acquired knowledge to use. The story goes something like this. …

The academic odd couple

About 15 years ago, working separately and without knowledge of each other's work, and with different initial motivations, two very different people began to question the state of biomedicine, specifically in the context of perhaps its most dramatic manifestation: critical illness requiring care in an intensive care unit; in other words, the sickest of the sick. Gary was and is a trauma surgeon and intensive care physician.

Yoram was and is a researcher, a biochemist and immunologist. They came from disparate training backgrounds, though they did, by coincidence, share the fact that they both spent time in the Chicago area and its northwestern exurbs. And, perhaps as a serendipitous harbinger of things to come, Gary knew Yoram's father, who had been a prominent surgeon in the Chicago area. They found out about this interesting coincidence when they met, quite by chance, at the Shock Society meeting in Big Sky, Montana, in 2002.

Gary was outside "The System," worked completely on his own between taking care of gunshot wounds at Cook County Hospital, on the fringe and unfunded, but already having published a paper on computational modeling in acute illness. Yoram was within "The System" but in his own way an outsider from within, part of a nascent interdisciplinary team that had just gotten its first major grant and started a biosimulation company, but with a focus and perspective well outside the mainstream.

This pair of academic oddities had absolutely no reason to ever meet, to find common ground, to like each other, much less to collaborate. But they had lunch together, by total coincidence, at that meeting in the May snow of Montana in 2002, and friendship was started.

An idea is born: Translational Systems Biology

Over time, the germ of an idea grew out many late-night discussions: what was it that we were doing that resonated with individual people but was distrusted, misunderstood, or ignored by so many others? Were the problems, deficiencies, successes and failures they observed specific to critical illness or a more general phenomenon?

What is Translational Science?

The exact definition of Translational Research is an ongoing subject of conversation. However, since the National Institutes of Health (NIH) is the preeminent funding body for biomedical research, the starting point for these conversations is often the definition in the call for its Institutional Clinical and Translational Science Award:

Translational research includes two areas of translation. One is the process of applying discoveries generated during research in the laboratory, and in preclinical studies, to the development of trials and studies in humans. The second area of translation concerns research aimed at enhancing the adoption of best practices in the community. Cost-effectiveness of prevention and treatment strategies is also an important part of translational science.

Translational Systems Biology was the product of these discussions. Translational Systems Biology is the application of advanced computer modeling and simulation based on biological knowledge and engineering principles, with the explicit goal of targeting clinical situations. While knowing how biological systems work at a fundamental level is critical to improving human health, Translational Systems Biology focuses on enhancing our ability to utilize that knowledge and bring the bench to the bedside in a patient-centric way.

Translational Systems Biology was a reaction to the "Translational Dilemma": the inability to translate basic mechanistic knowledge into clinically effective therapeutics. In short, the Translational Dilemma is why we don't have better drugs: we know more and more about the fundamental biology underlying disease, but we can't effectively turn that knowledge into better therapeutics. Unfortunately, simply gaining more knowledge doesn't seem to be the answer. The Translational Dilemma appears to be cropping up more and more frequently: paradoxically, our seemingly ever-greater molecular understanding of disease processes somehow makes trying to manipulate those processes ever more intractable. This is really troubling, since we have previously relied upon Science to help us out; now the very process of modern science appears to be only compounding the problem. Our diagnostic instincts tell us that this current situation is not sustainable, and therefore that we need to step back and reassess the fundamentals of the scientific process as an initial step towards identifying where and how the process can be augmented by technology.

The Scientific Cycle (and it is most certainly a cycle) involves iterating through the steps of Observation ("What do we observe?), Hypothesis Generation ("What are the patterns in what we observe?) and Hypothesis Testing ("How do those patterns arise?"). Technology has substantially enhanced several aspects of this cycle. We have better means of measurement and observation, allowing us to peer into biological systems to find unprecedented detail. And the advent of Big Data – i.e., the use of large-scale computation to analyze the huge amount of data now able to be collected – has allowed us to find patterns within our observations heretofore unrecognized.

<strong>Knowledge Ecology for the Future. </strong>A proposed future structure of the biomedical research community, implementing the computationally augmented Scientific Cycle at a community-wide level. The key principle depicted here is the transparency in the representation, generation, and utilization of mechanistic knowledge as encapsulated in computational models. The freely and openly available biomedical knowledge allows for distributed investigation, aggregation of disparate knowledge and the determination of the "fitness" of hypotheses based on their use and persistence within the research community. This research community structure incorporates evolutionary principles as applied to pieces of knowledge in a robust and scalable fashion. (Source: This image, which is Figure 5.1.2 in Vodovotz and An's new book Translational Systems Biology: Concepts and Practice for the Future of Biomedical Research, was reprinted with permission from Science Translational Medicine, where it appeared in An's 2010 article "Closing the scientific loop: Bridging correlation and causality in the petaflop age."But just knowing that pattern is there doesn't tell us why that pattern arises – and if we want to propose interventions we need to know how the system works. After all, knowing someone has cancer or how long she has left isn't enough; we want to be able to fix her. But the means of doing the necessary experiments to tease out the multiplicity of candidate "hows?" remains embedded in labs relying upon manual labor: the legions of techs, students and junior researchers who are foot soldiers of biomedical research.  However hard they and their guiding investigators work, they just cannot keep up with the technologically augmented output coming from the other aspects of the Scientific Cycle.

Therefore, our diagnostic approach has led us to the recognition that the Translational Dilemma is due to an imbalance of the Scientific Cycle, a bottleneck at the process point of evaluating our answers to the question "How does this system work?" – that is, the experimental testing of hypotheses. Ergo, the solution to the Translational Dilemma is to bring the advances of technological innovation to this choke point; hence the development of Translational Systems Biology.

[pullquote align="right"]"In short, we are providing a means of in silico testing of ideas and hypotheses, where electrons are faster and cheaper than cells (in vitro) and mice (in vivo)."[/pullquote]

We have envisioned Translational Systems Biology as a rational, systems engineering-oriented, computationally-based investigatory framework that can integrate data derived from basic biology experiments with pre-clinical studies and clinical studies, and ultimately lead to the development of strategies for rational manipulation of complex disease processes. In short, we are providing a means of in silico testing of ideas and hypotheses, where electrons are faster and cheaper than cells (in vitro) and mice (in vivo).

The methods of Translational Systems Biology are drawn from the most advanced engineering methods available, ones that have been applied to aerospace, computer chip design, industrial scale workflows, essentially everything having to do with the physical objects and material world around us, but tailored to the complexities of biology, where we are still (and will likely perpetually remain) in the discovery phase of understanding. The framework of Translational Systems Biology will hopefully allow us to accelerate the Scientific Cycle as applied to the biomedical arena, speeding the ability to evaluate promising ideas, and cast away and reduce time wasted on unproductive ones.

The appeal of this approach is that it takes what we already do – the Scientific Cycle that has successfully served us for over 500 years since the advent of the Scientific Age of Newton and Bacon – and updates it to the 21st Century and beyond by leveraging that sector of human knowledge with the most rapid, paradigm-shifting growth in the last 25 years: computation.

In our book, Translational Systems Biology: Concepts and Practice for the Future of Biomedical Research, we expand on these ideas. We show how, over nearly 15 years of work, we have applied computational modeling to the arena of acute inflammation and critical illness. We show key principles of Translational Systems Biology, including simulating clinical trials (see articles in Related Resources) and creating patient-specific computational models. We lay out a framework in which emerging infectious diseases, rising health care costs, scientific fraud, researcher burnout, failed clinical trials, and Big Data in Medicine emerge as logical consequences of the currently dysfunctional Scientific Cycle. Finally, we lay out the case for why this system needs to be disrupted in the same way as the high technology field – and perhaps by the same companies.


Watch a video of the authors

Drs. Yoram Vodovotz and Gary An talk about the state of biomedical research at the Harvard Coop in Cambridge, Massachusetts.


Related Resources

comments powered by Disqus

Share story:  

Related Stories