Report to the U.S. National Science Foundation (NSF) on the Outcomes and Consensus Recommendations of the NSF-sponsored International Workshop, February 2012, in Florence, Italy
Electronic publication of U.S. National Science Foundation's Industry-University Cooperative Research Center for Advanced Knowledge Enablement
33 multi-disciplinary organizations including
healthcare services, universities, public sector R&D,
and private sector information technology companies
from 11 countries and 8 US states
New directions needed for knowledge mining
and bioinformatics tools to impact patient care
i Compress translational timeframe
i Crack the economic code of Personalized Medicine
3 calls to action:
4FILL the translational white-spaces
4INNOVATE business models
The impetus for writing this book was inspired by a two-day NSF international Workshop on Knowledge Mining and Bioinformatics Techniques to Advance Personalized Diagnostics and Therapeutics that was moderated by Dr. Ron Ribitzky in Florence, Italy in 2012 (http://hit.fiu.edu/W/ and http://HIT.FIU.edu/W/pre-report.pdf). It is founded on the workshop deliberations and subsequent work of the moderator, members of the workshop’s scientific steering committee, and participants including chapter submissions on selected topics.
The content of this book is based in part upon work supported by the National Science Foundation under Grant OISE-1157372 “Knowledge Mining and Bioinformatics techniques to Advance Personalized Diagnostics and Therapeutics,” as well as by NSF grants Nos. I/UCRC IIP-1338922, AIR IIP-1237818, SBIR IIP-1330943, III-Large IIS-1213026, MRI CNS-0821345, MRI CNS-1126619, CREST HRD-0833093, I/UCRC IIP-0829576, MRI CNS-0959985, FRP IIP-1230661, SBIR IIP-1058428, SBIR IIP-1026265, SBIR IIP-1058606, SBIR IIP-1127251, SBIR IIP-1127412, SBIR IIP-1118610, SBIR IIP-1230265, SBIR IIP-1256641, I/UCRC IIP-0934364., including funds provisioned therethrough by the U.S. National Institute of Standards and Technology (NIST).
We are grateful for the guidance and support to the NSF Program Directors Dr. Carleen Maitland, Dr. Rita Virginia Rodriguez, Dr. Rathindra (Babu) DasGupta, Dr. Victor Santiago, Dr. Glenn Larsen, Dr. Larry Hornak, Dr. Barbara Kenny, Dr. Alex Schwarzkopf, Dr. Shashank Priya and Dr. Maria Zemankova, and the NIST Program Director Mary Brady.
This work was facilitated by the National Science Foundation's Industry-University Cooperative Research Center for Advanced Knowledge Enablement at Florida International and Atlantic Universities and Dubna International University (Russia) (I/UCRC-CAKE, http://CAKE.fiu.edu – Directors: Naphtali Rishe, Borko Furht, and Evgenia Cheremisina); NSF I/UCRC Center for Hybrid Multicore Productivity Research (http://chmpr.umbc.edu - Directors: Yelena Yesha and Milton Halem); NSF Center of Research Excellence in Science and Technology at FIU (http://CREST.fiu.edu – Director: Naphtali Rishe); Galil Center for Medical Informatics, TeleMedicine and Personalized Medicine, Haifa, Israel (www.GalilCenter.org.il – Founder: Uzia Galil, Director: Eddy Karnieli); the Technion - Israel Institute of Technology (http://www1.technion.ac.il/en); Rambam Medical Center (www.rambam.org.il); Up Close and Personalized 2012 Personalized Medicine Conferences (UPCP; http://upcp.org/ - Chairman: Eddy Karnieli) and updates at the 2013 and 2014 conferences; Paragon Conventions (www.paragon-conventions.com); US-Israel Science and Technology Foundation (USISTF) (www.USISTF.org - Executive Director: Ann Liebschutz); and R&D Ribitzky (www.RDRibitzky.com - CEO: Ron Ribitzky, M.D.).
Disclaimer of Endorsement:
Any opinions, findings, and conclusions or recommendations expressed in this book are those of the author(s) and do not necessarily reflect the views of NSF.
Reference herein to any specific commercial or public-sector product, process, or service by trade name, trademark, manufacturer, or otherwise is intended to provide an example and does not constitute or imply its endorsement, recommendation, or favoring by the authors, editors, and other individuals and organizations that are associated with this book (collectively, the ‘Book’).
Lack of reference herein to any other specific commercial or public-sector product, process, or service by trade name, trademark, manufacturer, or otherwise is not intended nor constitutes or implies unfavorable opinion, respectively, of this Book
Disclaimer of Hyperlinks:
The appearance of external hyperlinks in this Book does not constitute endorsement of the linked web sites, or the information, products or services contained therein.
Disclaimer of Liability:
This Book does not make any warranty, express or implied, including the
warranties of merchantability and fitness for a particular purpose, or assumes
any legal liability or responsibility for the accuracy, completeness, or
usefulness of any information, apparatus, product, or process disclosed, or represents
that its use would not infringe privately owned rights.
Contemplating and exploring the next wave of compelling problems that are worth pursuing on a global-scale is exciting indeed. Yet this book takes it further.
A rapidly evolving matter that is subject to changing economic conditions, scientific discovery, and clinical practice patterns, we intend to continue refreshing the content of this digital publication. The thoughts expressed in this book were inspired by an international multi-disciplinary workshop that was funded by the US National Science Foundation (NSF), and the US Israel Science and Technology Foundation (USISTF).
This book is intended for worldwide audience of policy makers, investors, program leaders, and scientists from the public and the private sectors.
The public at large is kindly encouraged to explore certain fundamental sections that are not ‘rocket science’. By this we mean that they do not require medical, scientific or technical knowledge to understand, deliberate on and take action.
And so this book informs the U.S. and other governments seeking to fund high impact research: whether in the form of an exclusive national pursuit, collaborative multinational endeavor, or global partnership among the public sector, private sector, and academia.
The workshop included four scientific sessions focusing on the policy implications of implementing personalized diagnostics and therapeutics based on big data analytics, the technological challenges facing computer scientists and physicians in creating useable systems, the challenges in utilizing big data analytics to predict future health outcomes, and the needs of clinicians in utilizing in their practices decision support systems based on big data analytics.
I greatly enjoyed facilitating this workshop using a technique based on Dr. Edward de Bono‘s theory ‘Six Thinking Hats'. My first encounter with this technique was at Intel Innovation Lab in Ireland a few years ago. It has proven useful and exciting again to elicit insightful thoughts and perspectives from this multi-disciplinary international forum. Examples are ‘Personalized medicine is ready for prime time. Why?..’, ‘Big Data is…’, ‘Predictive analytics projects will fail because…’, etc. At times through deep-thoughts, and at other by witty remarks, the use of this technique led to rich conversations and valuable debate that inspired, and served the foundation for this book.
The workshop reached a broad-based consensus on new directions for knowledge mining and bioinformatics tools to impact patient care; as well as strategic, proactive, and preventive health and wellness decisions here and now.
A multi-faceted, grand-challenge undertaking, the highlights included a call-to-action for technological breakthroughs to fill the growing ‘translational white spaces’ among the many scientific and clinical disciplines throughout the personalized medicine cycle up to end-user clinicians, patients, and consumers; innovative business models to accelerate the reduction of new discoveries along that cycle to practice; and policies that facilitate both.
Following the workshop we expanded the scope of the report out and supplemented it with chapters written by participants following the workshop. These chapters provide deep insights and specific real world key learnings from a range of personalized medicine initiatives already underway.
This effort was made possible by the personal dedication of over three dozen passionate practitioners, scientists and their respective organizations representing the many disciplines that converge on the single, globally shared mission: accelerate the translation of scientific discovery to making actionable clinical decisions.
I want to take this opportunity to extend my appreciation and gratitude to the members of the Scientific Steering Committee for the opportunity to take part in this important and fascinating event. A special ‘thank you’ to Ann Liebschutz and her team at USISTF which included Eve Copeland, Charlie Swartz, and Robert Brunson for their outstanding support and assistance in making this book happen; and to Karell Müller, U.S. National Science Foundation’s Industry/University Cooperative Research Center for Advanced Knowledge Enablement at Florida International University for working with me on the graphic design for the cover of this book.
To my wife Dafna and my children Romy, Laura, Tom, and Roy – thank you for being so supportive from inception to completion.
We call it ‘A-to-Z’.
Ron Ribitzky, M.D.
CEO, R&D Ribitzky
A note from the Scientific Steering Committee
The world has seen the human genome fully decoded by an international team of scientists after more than a decade of work to being available to scientists within days or hours. The question now is how will we use the wealth of information available to us through our newly understood genomic data and, further, given our massive computing power, can we merge this information with all patient’s health data, compared with like patients and exponentially growing medical knowledge in order to better diagnose and offer therapeutics?
With the generous support of the U.S. National Science Foundation and our respective organizations we convened top scientists, practitioners, and industry leaders from multiple disciplines to take a hard look at what needs to be developed to help the industry grow; and chart a path to compress the cycle time and economics from emerging technologies and methodologies to deployable high-impact, scientifically-sound industrial-grade solutions.
We tasked this distinguished forum to factor-in regulatory, legal, technology infrastructure and other drivers and barriers to developing practical and achievable personalized medicine solutions that can be deployed around the world. This includes the developed and emerging economy countries alike.
We have requested Dr. Ron Ribitzky, an independent subject matter expert to facilitate this two-day workshop: from framing and focusing the questions to driving an open and lively debate, and wrapping it all up with scholarly peer-reviewed publication. The Committee would like to extend a very sincere Thank You to Dr. Ribitzky for his amazing work. We hold skills in the highest regards, as his extremely competent facilitation skills were instrumental in producing the high caliber output.
We need more scientific and outcome oriented workshops of this kind in an ongoing effort to bring again computer scientists and clinicians together and lay the groundwork for the future of medicine and translating this foundation into provisioning of the relevant information to the general practitioner treating the patient.
Prof. Naphtali Rishe
Prof. Yelena Yesha
Prof. Eddy Karnieli
Ann Liebschutz, J.D.
Workshop Chair and NSF Principal Investigator
Program Steering Committee
Table of Content
Call to Action
1. The ultimate goal of Personalized Medicine is making actionable decisions that matter about one’s health or wellness, here and now
2. We envision the emergence of new markets and new industries with high-value breakthrough innovation and cross disciplinary pollination
3. Look for key learnings from Intel’s ‘Copy Exactly’ and ‘Tick Tock’ innovation and production model for next generation breakthroughs in personalized medicine
1. Fill high-value white-spaces in translational research and personalized medicine
2. Innovate business models aimed at making the practice of personalized medicine a commodity
3. Facilitate both through legal and regulatory frameworks, and public-private partnerships
technological, cloud, policy, legal, commercialization, economic value model, adoption, innovation
Making actionable decisions about one’s health or wellness, here and now, is the ultimate purpose of personalized medicine.
It was the theme of the National Science Foundation’s direction-setting International Workshop on Knowledge Mining and Bio-informatics Techniques to Advance Personalized Diagnostics and Therapeutics in Florence, Italy 2012 that inspired, and continues through this book.
And so we have set out to discover future directions for high-value breakthrough research in bio-informatics that would influence concrete actions that matter in making these decisions.
We call it filling the white space of translational research and personalized medicine.
Contemplating and exploring the next wave of compelling problems that are worth pursuing on a global-scale is exciting indeed. Yet this book takes it further. We seek to accelerate emerging and ignite new technologies to match these problems.
We envision the emergence of new markets and new industries that would thrive on filling the white space in personalized medicine and translational research with high-value breakthrough innovation and cross-disciplinary pollination.
This book aims to inform public and private sector organizations seeking to fund high impact research: whether in the form of an exclusive national pursuit, collaborative multinational endeavor, or global partnership among the public sector, private sector, and academia. And so, each chapter begins with a summary of key points (take-aways) and a call to action.
Where one begins? What’s the roadmap?
Personalized medicine is evolving through paths of micro tipping-points. Nevertheless, it takes too long and costs too much to transform scientific discovery to meaningful clinical actions. In the chapter on adoption, we offer a dozen strategic approaches to accelerate application of discovery to practice aimed at cutting time lags 2-3X in 2 to 3 cycles.
Consider this: we have massive computing power that continues to grow. Analytics and knowledge management paradigms continue to evolve rapidly. Social-media grows exponentially on a global-scale, threading through communities of scientists, practitioners, policy makers, investors, entrepreneurs, and - of course, the general public. The network effect of these multi-dimensional connections and diverse content is colossal.
What does it take to merge all these with one’s illness, wellness, bio-markers, and environmental data in order to better predict, prevent, diagnose and treat his or her medical problem in a timely and economically affordable manner? Joshi, Joshi, Yesha, and Yesha offer an approach to construct a formidable policy based information technology infrastructure to do just that.
The question now is how will we use the wealth of information available to us through our newly understood genomic data and, further, given our massive computing power, can we merge this information with all one’s health data, compared with like patients and with exponentially growing medical knowledge in order to better diagnose and offer therapeutics?
Hsu, Tran, and Linninger inform us that we begin to master the capabilities to digitally reconstruct one’s organs with such high degree of precision previously limited to manufacturing; and apply predictive analytics to that functional expression for personalized molecular therapy.
Does it mean that informatics is no longer the best term to describe the use of information technology in translational research; and would it make more sense to extend the use of ‘Infomics’, in line with genomics and proteomics?
And because personalized medicine is entering the manufacturing space, albeit in the virtual digital form, could or should we look for key learnings such as ‘Copy Exactly’ that is key to Intel’s decades long technological and commercial success and ‘Tick Tock’ innovation and production model for next generation breakthroughs in our field? In like manner, virtual organ reconstruction and disease modeling for personalized medicine would be performed with ‘Reconstruct Exactly’ zeal and practice in mind.
How can that be achieved as information technology challenges grow consequent to the upward pace, volume, and diversity of scientific discovery in life sciences continue throughout the translational medicine stack; episodic and longitudinal patient information accumulate in corporate information systems of healthcare providers and insurers; and individuals stream personal and wellness information up and down clouds and through social networks.
And so, Mennel proposes to form a “consortium of physicians, basic scientists, institutions, and companies to limit their investigation to the studies that will likely answer the most important questions and not try to answer every question.”
Yet many factors are at play in “What is the best question to pursue in a trial?” Campbell et. al. offer “Scalable methodologies for distributed development of logic-based convergent medical terminology”. Would such consensus-building techniques help converge the widely diverse stakeholders of personalized medicine?
How could that shift the balance where over 200 other doctors are involved in treating the Medicare patients of an average primary care physician; and one-third of health care expenditures does not improve health?
What are the shortcomings of contemporary business model thinking? And what business model innovation may emerge that would make personalized medicine a commodity?
We offer a domain model to help the different stakeholders frame and focus such questions; and a value model to facilitate making tough choices among promising personalized medicine projects in light of limited resources. Liebermann, Klang, Recanati, and Balicer share with us what may seem counter-intuitive case studies of national scale adoption of personalized medicine by a managed care organization.
An insightful account of Mayo Clinic Vs. Prometheus wraps up this book. This landmark case of intellectual property pertaining to personalized medicine divided the industry. Avoiding being prescriptive, our hope is that it will enrich our readers’ positions and decisions as to whether or not to patent a new discovery in this field.
Author: Ron Ribitzky, M.D. Contact: Ron@RDRibitzky.com
Call to Action
1. We describe the personalized medicine domain model to frame and focus the exchange of ideas among the multiple stakeholders aimed at accelerating the benefits of personalized medicine
2. The personalized medicine domain model addresses disease lifecycle and translational research
3. The personalized medicine domain model facilitates exploration of bench-to-bed and bed-to-bench program and project opportunities
4. We offer a framework to explore the impact of strategy and action in the practice of personalized medicine
5. Growing trustworthiness challenges make the pursuit of evidence in personalized medicine a monumental effort that amounts to a white space issue
1. Model and measure value that is based on strength of evidence at each major aspect of the disease lifecycle and translational research leading to actionable personalized medicine decisions
2. Fund white space R&D to tackle the growing trustworthiness problem in translational research and personalized medicine
technological, cloud, policy, legal, commercialization, economic value model, adoption, innovation
Personalized medicine is a complicated high-stake field. Multiple disciplines are at play: from clinical to life sciences, information technology, business, legal, and regulatory, to name a few. Terms and concepts commonly used in one discipline may be vague or simply obscure to professionals in another. And the pace of discovery seems to accelerate exponentially while transforming it to clinical action may still take over a decade and cost over $1B.
And so we propose a value-based conceptual domain model to frame and focus the exchange of ideas among the multiple stakeholders aimed at accelerating the benefits of personalized medicine.
At Risk phase is when a person may have a tendency to be sick, yet no evidence is found that a disease process has begun.
The beginning of a disease process marks the transition from At-Risk to Preclinical Progression phase. The patient may not feel or otherwise realize that he or she is sick, and pathology may go undetected by means available to them.
During the Disease Initiation phase, patients may begin feeling that something is wrong with them, or early signs of disease may be objectively detected.
Disease Progression is when the symptoms and signs of illness are quite obvious. Certain diseases may progress to an early chronic phase. The duration of their illness is prolonged, and it may be more difficult to be cured.
Although not every illness ends up being chronic, some may deteriorate into an Irreversible Damage phase.
The disease lifecycle can be illustrated in a process flow diagram:
For sake of simplicity, let’s assume that genes constitute the foundation of personalized medicine; and that genotype and phenotype are the outermost edges of the personalized medicine domain. What is left for us to explore then are the transitional domains in between: the molecular and cellular.
Respectively, let’s agree that the molecular domain is essentially the expression of one’s genome in molecular terms; and that the cellular domain is essentially one’s expression of the molecular play in terms of tissues and organs.
Because our concern is one’s illness as well as wellness, we’d need to complement normal expressions with expressions of abnormality. What comes to mind is the use of disease models as a descriptive domain in our conceptual framework.
Because complex mechanisms of expression are at play (some are quite convoluted), let’s refer to them as pathways.
Nevertheless, we know that environmental factors may influence the structural elements and their properties, eventually changing the original (or personalized) script quite substantially. So let’s include Environmental Impact domain in our framework.
Yet informatics professionals may more intuitively relate to a stack model of the same - the equivalent of an architecture stack, a fundamental model widely used by information technologists. Similarly, the bidirectional arrows illustrate the equivalent of round-trip flow of information and knowledge required to continuously drive our understanding of translational medicine.
Let’s begin constructing the domain model of personalized medicine by layering the disease lifecycle dimension on top. The equivalent of top-down approach, let’s call it the Bed-to-Bench methodology. We are mindful that ‘Bed’ may suggest a progressive state of disease that bounds the patient to bed. However, for our purpose here, we borrow the widely known concept Bed-to-Bench to imply that clinical considerations drive scientific pursuit.
Inherent to this approach, of course, is the presumption that an individual may suffer from a single disease. It is often not the case. Nevertheless, at this stage of orienting ourselves to the conceptual framework, we will keep it simple, i.e. to a single disease. We address the impact of co-morbidity on personalized medicine elsewhere in this book.
Let’s now bring in the translational research dimension to inform us about the current state of the industry pertaining to the disease we are exploring. In this we mean the pipeline of scientific discovery, and opportunities that it may provide us. The equivalent of bottom-up approach, let’s borrow the other widely known concept Bench-to-Bed methodology to describe it. In case you are not familiar with this term. Bench-to-Bed implies that scientific discovery may inform clinicians about new options available for their patients.
Putting the two together will look like this:
Combining disease lifecycle and translational research models provide us the equivalent of a map for treasure hunt as well as the beginning of a story board. Both are useful for identifying high-value targets for white-space research and development and roadmaps.
The actionable dimension of our framework seeks to map out the key actionable themes in reference to the disease lifecycle.
Nevertheless, to keep our framework simple, we propose three actionable themes that – either practiced singularly or together, may have substantial impact on driving personalized medicine and translational research.
Reactive is taking action based on the recognition that illness is in process. For some, reactive medicine marks the legacy of clinical practice up until recently, and may continue to be commonplace so long as personalized medicine may not be widely adopted.
Preventive is the active pursuit aimed at avoiding the occurrence of undesirable condition.
Predictive is premised on one’s ability to determine with reasonable evidence that certain condition may or may not occur, further driving reactive and/or preventive action.
Each of these themes may be further examined along a Strategic dimension which extends the timeframe of their respective impact.
The evidence dimension seeks to map out key elements that actions should rely on in personalized medicine. Actions may range from setting direction for research, embarking on the development of next generation medical devices, implementing new clinical protocols, adapting policies, defining regulatory requirements, etc.
Outcome expresses the end-result of causality throughout the descriptive dimension of our framework. Let’s agree that outcomes may be expressed in qualitative and quantitative terms. Be they as they may, for outcomes to be considered they should be measurable.
Nevertheless, white space in personalized medicine and translational research implies that still there is much to be discovered. Therefore, let’s also agree that strength of evidence may be expressed in semi-quantitative terms (such as high, medium, low). The most extreme case would be ‘none’ (respectively, ‘boundary condition’ and ‘null’). This is to say that an assumption of causality or outcome cannot be objectively substantiated.
We offer time-based model and stack model graphics of Strength of Evidence (SOE).
In “Sloppy Science and Useless Information”, Burrill points out that the validity hence trust-worthiness of published scientific discovery has become a major concern:
§ 15X retraction rate growth on 44% increase in the number of publications in research journal since 2001 (Source: Wall Street Journal)
§ 15.4X growth in retracted articles among 16,000 peer reviewed journals between 2001 and 2010 (Source: Thomson Reuters Web of Science )
§ 7X growth of retractions related to fraud in medicine and biology studies published in the Journal of Medical Ethics during a 5-year period between 2004 and 2009
§ 2X growth of retractions related to errors in medicine and biology studies published in the Journal of Medical Ethics during a 5-year period between 2004 and 2009
Authors: Ron Ribitzky, M.D. and Dr. Joel Saltz, MD, PhD Contact: Ron@RDRibitzky.com
Call to Action
1. An intuitively obvious concept, personalized medicine is inconsistently defined
2. A person’s biology is a system of expressions that derive from their individual information
3. We offer taxonomy of concepts related to the term “Personalized Medicine”, and a pragmatic framework to facilitate the application of these concepts for particular purposes and contexts
1. Establish open-source semantic network of personalized medicine content in partnership between the public and private sector
Definition of personalized medicine, taxonomy, framework, technology
Reaching consensus over what is Personalized Medicine turned out to be a non-obvious undertaking. At the outset, a fundamental question dominated the effort of this multi-disciplinary, international forum to reach common grounds for the upcoming exploration:
What is medicine that is not personalized?
Except for population health, a nuance follow up question emerges: Is there a line between personalized medicine and medicine that is not personalized?
Clinicians wonder what the fuss is about, indicating that for them the practice of medicine was always personal – exclusively dedicating their passion, knowledge and skills to each patient individually, one person, one encounter at a time.
Yet certain scientists hold the opinion that Personalized Medicine is a relatively new, emerging field synonymous with nothing else but genomics. Some felt that proteomics might be considered too for this discipline-grounded definition of Personalized Medicine.
Some advocate a completely different approach to defining Personalized Medicine, recognizing that the subject matter is indeed very broad, very deep, and multi dimensional. Aside from academics, one may argue that it may be impractical to attempt reaching a singular definition. Instead, one may try sub-typing Personalized Medicine so that we can have Personalized Medicine of Type A, B, C, etc. The potential value of sub-typing Personalized Medicine lies in the ability to provide precise definitions of smaller scope which would drive clarity rather than ending up with a too high level, vague one.
Or is it?
The US National Cancer Institute (NCI) of the National Institutes Health (NIH) defines it as “a form of medicine that uses information about a person’s genes, proteins, and environment to prevent, diagnose, and treat disease.” The U.S. National Library of Medicine® defines Personalized Medicine as “an emerging practice of medicine that uses an individual's genetic profile to guide decisions made in regard to the prevention, diagnosis, and treatment of disease.”
Adopted by the personalized Medicine Coalition, the US President’s Council on Advisors on Science and Technology in 2008 refers to Personalized Medicine as the “tailoring of medical treatment to the individual characteristics of each patient…to classify individuals into subpopulations that differ in their susceptibility to a particular disease or their response to a specific treatment. Preventative or therapeutic interventions can then be concentrated on those who will benefit, sparing expense and side effects for those who will not.” 
Eddie Blair of the UK-based Integrated Medicines suggested that Personalized Medicine means the right medicine for the right patient for the right disease at the right time and right dose for the right response and the right price. 
Seeking to mediate the emerging spectrum of existing and potentially new definitions, experts in life sciences informatics offered the fluid approach: instead of trying to define Personalized Medicine in a binary, singular and exclusive fashion it may be useful to characterize it by the type of data that may be involved in a particular real world situation – i.e. the qualitative dimension; and by the volume of data that may be required – i.e. the quantitative dimension, pointing at the rising ‘Big Data’ field.
The notion of fluid approach stemmed from the forum’s consensus that fundamentally, a person’s biology is a system of expressions that derive from their individual information; and that Personalized Medicine is the practice of actionable reasoning about it. Pathways of translation and expression, external environmental factors, and homeostasis add up to fulfill the needs of contemporary discourse about Personalized Medicine.
The ever growing pace of new discovery and deeper sub-specialization make actionable reasoning exponentially daunting challenge for the entire ecosystem: from scientists to clinicians, policy makers and private sector, and the public at large – i.e. the ‘Person’ in ‘Personalized Medicine’ that is the grounding focus of this forum.
Osler’s 100+ years old concept of medicine as the practice of comprehensive and careful observations is a useful baseline: “the integration of scholarship with patient care, together with the science and art of medicine... concerned with the ideals of medicine as with its science and knowledge”.
Citing John Keats’s reflections in 1817, Dr. Nuland points out that medical education was short back then. “Although the examinations were difficult, there was little of real usefulness to learn… patient care was conducted in a pervasive atmosphere of inexactness.”
Providing pragmatic perspectives that are founded on philosophically insightful considerations, Dr. Nuland discusses the context for and role of clinical judgment: “To become comfortable with uncertainty is one of the primary goals in the training of a physician… clinical decision making is the realization that, perforce, it must always be accomplished in the face of incomplete and largely ambiguous information.” 
Omics are commonly referred to as rapidly evolving fields of study that range from the structure and behavior of genes (Genomics), proteins (Proteomics), metabolites (Metabolomics) and other. These fields seek to develop omics-based tests and methods to accelerate progress of, widen access to, and reduce cost of wellness programs, disease prevention, and patient care.
And so, Burrill contemplates whether personalized medicine “[is] a surrogate for molecular diagnostics?”.
Call to Action
1. We describe strategic actions and roadmap to accelerate application of discovery to practice aimed at cutting time lags 2-3X in 2 to 3 cycles
2. Engaging patients in personalized medicine is essential for R&D and its adoption
3. Personalized medicine is evolutionary in certain aspects, and revolutionary in other
4. Personalized medicine is evolving through paths of micro tipping-points
5. Personalized medicine is a discipline of disciplines driving the creation of new ones along the way
6. Clinical informatics is a US board-certified subspecialty of primary care physicians
1. Investigate and develop strategies to engage patients, clinicians, and scientists in collaborative effort to influence research directions and adoption of personalized medicine
2. Fund white space R&D to accelerate knowledge and information transfer between one translational medicine domain to the other
Innovation, cloud, policy, legal, commercialization, economic value model, adoption
As scientifically fascinating a discussion may be about genomics and proteomics, pharma and biotech, big data and super computers, personalized medicine is essentially about patients: whether one knows that they are ill, or possess certain markers that are strongly suggestive of illness.
Personalized medicine is a global domain that traverses national and political boundaries – from formal collaborative research and development efforts to informal exchange of information, knowledge, and experience among professionals and consumers.
Clinicians need to understand what information they’re getting, and they need to get it fast – typically within 10 minutes or less, because of the limited time they have for the encounter with the patient. Usability of electronic health records and interoperability of health information systems are formidable barriers to achieve this goal.
Ironically, exponential growth of scientific and clinical discovery makes it no less exponentially difficult to put together with sufficient evidence to make it actionable for clinicians.
Discovery runs faster around the world than improvements in electronic health records (EHRs), electronic medical records (EMRs), healthcare information systems (HIS), health information exchanges (HIEs), interoperability standards, and laws and regulations governing clinicians access to information about their patients.
Personalized medicine is evolutionary in certain aspects, and revolutionary in other.
That personalized medicine is an evolutionary phenomenon founded on decades if not centuries of just fundamentally right clinical practice was strongly held position mostly by clinicians. It is reinforced in the chapters about precision medicine and adoption of personalized medicine by non-profit HMO further in this book.
Nevertheless, personalized medicine is revolutionary as well, considering the exponential growth of highly specialized research projects by the many life sciences disciplines and sub-disciplines that are driven by and further drive progress in personalized medicine; exponential growth in volume and complexity of data, information, and knowledge directly and indirectly impacting personalized medicine; and a rapid stream of new information technology components and capabilities that drive new usage models never before possible in science, business, and social life.
And that revolution is disruptive.
We can cure diseases we were unable to cure before; predict diseases we could not foresee before; prevent diseases we could not prevent before; be proactive like never before; and provide treatments not available before.
Conceivably, clinicians are able to practice in much higher specificity and efficacy more easily and faster, relying on complex and complicated information management and analytic processes and techniques. While the practice of personalized medicine may or may not be revolutionary by itself, the outcomes and impact of personalized medicine on the individual and others is. Certain chapters in this book describe specific real-world accomplishments along these lines.
The depth and breadth of understanding complex pathways and mechanisms of wellness and illness, as well as cure and prevention are by themselves revolutionary. From genomics up to translational omics, this disruption requires us to think and act differently about personalized medicine. It requires us to explore, develop, and evaluate new approaches to model and measure the multi-faceted value of new opportunities made available by personalized medicine to impact quality of life and economics.
Furthermore, the far-reaching innovative thinking and pace of innovative technologies that drive all this are fundamentally revolutionary. And so, there are the perpetual cycles of evolution and discovery, revolution and sharp turns in practice and outcomes.
This is why we call it the paths of micro tipping-points.
Patients and the public at large play a key role in growing market demand for personalized medicine. There are clear or very obvious benefits to patients and other who may not become patients thanks to personalized medicine. Health 2.0 and social media are the driving forces in making the public increasingly educated, informed, and value-driven consumers.
Standardized one-size-fits-all protocol-guided medicine can be harmful to individuals who do not perfectly fit the persona of human subjects in controlled clinical trials.
Toxicity, side effects, comorbidity, further deterioration due to delayed diagnosis and treatment, and collateral adverse effects on social and economic well being of patients, their families, and their loved ones were cited as common undesired outcomes of population-based protocol guided care.
There’s enough knowledge right now to support starting the practice of personalized medicine and professionals willing and able to practice it.
A growing body of evidence suggests that the scientific foundation of certain discoveries pertaining to personalized medicine is sound and dependable to warrant action. Nevertheless, we discuss the growing trustworthiness challenge of scientific publications in the section on evidence in this book.
Personalized options to prevent, decelerate progression, and treat diseases are already available. Full human genome sequencing is offered commercially and its cost is going down. Actionable biomedical markers and respective protocols with great specificity and strong evidence of cause, effect, and outcome are commercially available, and new ones are in the pipeline.
Public and private sector collaboration around the world continues to drive wide accessibility to full human genome sequencing by making it less costly. With that, benefits to patients and those who may become ill, their families and loved ones, their medical providers, payers, governments, and society at large are obvious. The body of evidence of favorable, near and long-term clinical and economic outcomes of personalized medicine is growing.
The chapter about large scale adoption of personalized medicine by a non-profit HMO provides pragmatic real-world examples.
A growing number of personalized medicine elements such as diagnostic procedures for early detection and personalized best-fit therapeutic measures are commercially available, and more are in the pipeline, worldwide.
Enterprise-scale data warehouse and analytics that enabled the emergence and early adoption of personalized medicine are mature and main-stream. Big data analytics with broader reach and deeper capabilities enter the personalized medicine space in compelling fashion, driving innovation and providing opportunities for discovery like never before.
Disease-modeling complement the vast content in data warehouses and Big Data environments to provide near-real-world simulations and evaluation of actionable personalized medicine opportunities. Cloud-based computing and social media that connect professional and lay communities enable exponential growth in access to personalized medicine.
The chapters on information technology, cloud, and disease modeling further the discussion about these and related points.
Educating would-be physicians and nurses, and clinicians already in practice in the fundamentals of informatics is a significant factor determining wide spread adoption of personalized medicine. Understanding what are and what may be the questions in the practice of personalized medicine, and how information technology and informatics services can be used for that purpose: one encounter and one patient at a time are key to making personalized medicine the norm.
Generally speaking, primary care physicians and practitioners in the community are less ready to practice personalized medicine in real time than their fellow clinicians in academic medical centers and tertiary care organizations.
Seeking to address clinicians’ competency in informatics, the American Board of Medical Specialties in 2011 approved clinical informatics as a board-certified medical subspecialty of primary care physicians. AMA recognized that the practice of medicine is increasingly data-driven and dependent on information technology.
The role of Primary Care Physicians (PCPs) certified in clinical informatics includes:
§ Assessing the knowledge-based needs of health care professionals and patients
§ Characterizing, evaluating and refining clinical processes
§ Developing, implementing and refining clinical decision support systems
§ Leading or participating in the procurement, customization, development, implementation, management, evaluation and improvement of clinical information systems
This landmark move followed a six-year campaign led by the American Medical Informatics Association (AMIA). AMIA developed the core content of the clinical informatics subspecialty with support from a grant by the Robert Wood Johnson Foundation
The chapters on policy driven cloud based services for personalized medicine and patent considerations provide additional compelling insights on these topics.
Notwithstanding the many paths of micro tipping-points, paved by exponential growth in pace and volume of new discovery, “Patients suffering from debilitating and life threatening diseases do not have the luxury to wait the 13 years it currently takes to translate new scientific discoveries into treatments that could save or improve the quality of their lives.” (Dr. Francis Collins, NIH Director). Estimates of that wait-time range from 13 years to 15 and 17.
Or is this hurry-up-and-wait problem rooted in exactly that: many paths of micro-tipping points and exponential growth in pace and volume of new discovery?
FasterCures, a private non-profit center for accelerating medical solutions attempts to speed up the time from discovery to patients by improving the medical research system.
Reducing the cycle time from discovery to practice can, and should be exponential rather than linear.
Yet exponential acceleration in personalized medicine requires addressing the whitespace that holds back progress from one domain of translational research to the other.
The NSF workshop forum reached consensus that the time lag from discovery to practice can be cut 2-3X over 2 to 3 cycles. It is achievable by executing on the following strategies and factoring synergy of programs, campaigns, and outcome measures.
§ Establish value modeling programs for translational research and personalized medicine initiatives
§ Develop and disseminate value models for personalized medicine that cater to the needs and concerns of consumers and patients, private sector investors, ecosystem players, and public sector funding
§ Develop economic, scientific, and professional incentive campaigns that reward acceleration of personalized medicine solutions
§ Use value models to focus the effort on value creation and set the stage to measure it…
Research and Development (R&D) strategy
§ Set contests for achieving these goals based on the value models and incentives campaigns
§ Measure success and elicit key feedback on what worked, what did not work, and how the next R&D cycle could be more productive (the equivalent of win/loss reviews in private sector marketing and sales)
§ Adapt powerful horizontal technologies and explore the fit of special-purpose technologies from other industries to the acceleration R&D agenda
§ Establish programs to measure the adoption rate of these new technologies and techniques
§ Exploit accelerator technologies and techniques and breakthrough innovation to design content repositories as loosely federated collections of discrete, semantically-enabled fine-grain content components capable of expressing their attributes, general behaviors, and specific interactions throughout the translational research and personalized medicine domains
§ Create and provide open access to massive repositories (physical and virtual clouds) of omics expressions with associated phenotypes and longitudinal clinical data
§ Accelerate the development and adaptation of disease models that exploit these powerful information technologies, techniques, and content
§ Leverage content exchanges to discover acceleration opportunities and assemble new tools, services, and solutions
§ Develop and execute global social media campaign to foster multidimensional cross-disciplinary connections that leverage the network-effect from discovery and innovation to implementation aimed at achieving the research, development, and business goals
§ Exploit the social media campaign to facilitate the equivalent of virtual personaliz ed medicine teams
§ Develop special-purpose personalized medicine education and training programs for consumers and professionals with major emphasis on contemporary informatics techniques
§ Measure the success of the social media campaigns in terms of content scope and scale of engagement
§ Exploit global content exchange to adapt and optimize regulatory strategies to N-of-1 type studies in extraordinary situations of human suffering and consumers choice
§ Leverage content strategy for the key components from multiple N-of-1s over time that may apply to larger populations
§ Adapt regulatory oversight for speed and utility of applying new discovery in clinical practice by optimizing clearance for informed risk and benefit decisions of patients, consumers, and clinicians
§ Measure the success of the regulatory strategy by Time To Market
Call to Action
1. For personalized medicine to become widely adopted, information technology should be capable of elegantly displaying clinical roadmaps constructed from insights that are discovered within enormous amounts of massively disjointed data in clinical encounter timeframes at a price of single-digit pay-per-use
2. Readiness of personalized medicine for prime time should be carefully evaluated based on venue, maturity of information systems available there, and practitioners’ proficiency in informatics.
3. Building and operating centralized Personalized Medicine Oriented Infrastructure is beyond the capability of any singular organization
4. Bringing the information technology infrastructure is predicated on crafting a methodical multi-tier service-level interoperability framework and roadmap
1. Fund R&D for search and discovery technologies that bridge across the white space in personalized medicine and translational research
technological, cloud, policy, legal, commercialization, economic value model, adoption, innovation
Technological innovation and scientific discovery abound, yet ironically their confluence at the point of making actionable personalized clinical decisions is far from commonplace widespread reality.
The issues we’re facing are not simply about tangible matters such as processing speeds, operating systems, applications, next generation wireless bandwidth, big data or clouds. Once these are accounted for, new challenges commonly referred to as soft issues are more difficult to overcome.
Tackling adoption as the product of usability, interoperability, and value of IT anywhere in the personalized medicine domain model we present at the beginning of this book is the equivalent of the perfect storm of personalized medicine informatics.
It is complicated indeed.
This chapter examines certain formidable challenges leading to the informatics gap of the kind of personalized medicine we discuss throughout this book, describes mega trends that should be factored in, and call to action to close that gap.
The chapters that follow expand on some of these topics through a combination of scientific projects underway, pragmatic real-world case studies, and key lessons.
To enable the widespread practice of the kind of Personalized Medicine we discuss throughout this book, technology should be capable of elegantly displaying medical roadmaps constructed from insights that are discovered within enormous amounts of massively disjointed data in clinical encounter timeframes at single-digit pay-per-use price.
Whether it happens in a legacy brick-and-mortar environment or in the increasingly prevalent virtual encounter, it means the capability to create compelling measurable value to the stakeholders that matter during the time that a person consults with someone or something about their health, wellness, or illness.
We call it informatics for personalized medicine at clinical-clock speed.
Are we there yet?...
Clinicians’ access to state of the art Electronic Medical Records (EMR) and clinical decision support systems varies greatly. Access to personalized medicine support services capable of evaluating one’s omics signature with correlative analytics at high strength of evidence is a formidable challenge to be reckoned with in figuring out whether personalized medicine is ready for prime time.
The Healthcare Information and Management Systems Society (HIMSS), publishes periodic updates of healthcare provider organizations’ maturity of Electronic Medical Record systems that can be useful to determine their readiness for personalized medicine.
The scope of infrastructure required for personalized medicine that this book calls for is more complex than conventional healthcare IT, and its scale is vast. It amounts to a white space, grand-challenge as we discuss throughout this book.
We call it the Personalized Medicine Oriented infrastructure (PMO-I). Respectively, we call the architecture that drives it the Personalized Medicine Oriented architecture (PMO-A).
Monumental effort is required to achieve optimization for usage models, performance, information models, databases, configuration lifecycle management, and cost.
Thinking through the kind of architecture, infrastructure, implementation, and operations that are needed for personalized medicine takes an impressive assembly of senior level professionals. For a start, such will include Solution Architects, Chief Technology Officers (CTOs), Chief Information Officers (CIOs), Chief Medical Information Officers (CMIOs), and Chief Information Security Officers (CISOs).
This highly talented group will have to factor in cross disciplinary collaboration frameworks and respective technologies such as the US National Cancer Institute's caBIG® (cancer Biomedical Informatics Grid®); i2b2 (Informatics for Integrating Biology and the Bedside); the Total Cancer Care Consortium at Moffit ; Scripps Translational Science Institute; The European Commission 7th Framework Programme’s CORDIS (Community Research and Development Information Service); UK National Health Services National Translational Research Partnerships; etc.
Also at play are rapid uptake and confluence of social media and mobility driving data, information, and communication to cross traditional boundaries between private, corporate, and research environments; exchanging and expressing facts, opinions, thoughts, and sentiment. This subject requires broad and deep discussion that are beyond the scope of this book.
We anticipate that the pace of change of social media and mobility will far exceed the speed of publishing and readership outreach of this book. Therefore, media other than hard copy publishing would better serve this topic and our intended audience.
The consumer-centric BlueButton method for health and wellness information exchange in the US is worth noting: from user’s experience having real time control over granting access to their health and wellness information at a push of a button, to the vast scale of adoption that is within reach, and the formidable public-sector marketing campaign to make that happen which is driven by the Office of the President of the Unites States.
The latter complements the US Government’s financial and political drive to accelerate adoption of Electronic Medical Record systems in a multi-year effort that became known as ‘Meaningful Use’; establish Health Information Exchanges (HIEs); and set up Health Insurance Exchanges.
Service-oriented approach such as contemplated, for example, by the Object Management Group’s Open Health Tools is inevitable to achieving personalized medicine oriented infrastructure capable of the following essentials:
§ Bringing in legacy environments
§ Bridging across disciplines, organization boundaries, and national borders; and -
§ Providing the kind of agility to address rapidly changing technological landscape
Semantic normalization is required to assure that terms, concepts, and contexts are fully understood and applied consistently throughout our personalized medicine domain model.
However, providing detailed strategy and technical information about the interoperability framework is beyond the scope of this book.
To illustrate how formidable PMO-I is, consider the reiterative ripple effect that a single new discovery may have on correlating cause and effect, contemplating the potential associations between and among markers and outcomes, and exploring the mechanisms that may explain them: from structural genomics to clinical outcomes.
Now repeat that with the next discovery and the one that follows. The challenges are exponentially compounding indeed.
Therefore, building and operating a centralized PMO-I is beyond the capability of any one organization, public or private – be it a national government or academic institution, global corporation or a start up. Extensive collaboration across diverse professional, scientific, and technological disciplines is therefore inevitable. The resulting cross-pollination that PMO-i fosters is a potent generator of superior high-value breakthrough innovation.
A computing model providing web-based software, middleware and computing resources on demand for patients and medical practitioners alike is presently the ultimate PMO-i. From ubiquitous, always-on provisioning of executable content and collaboration environment to near real time and predictive analytics, and from semantic orchestration to global localization, cloud-enabled POM-i services eliminate the need to replicate complex and prohibitively costly infrastructure components.
And so, cloud-enabled data services have the potential for making POM-i widely and easily accessible to the public at large, an everyday thing, for laymen and professionals alike. Commoditizing access to POM-i can drive return to adoption, and business model innovation that can fuel the white space R&D.
Conceptualizing the practice of medicine as a learning system is an emerging approach that may help tackle this grand challenge towards making Personalized Medicine commonplace. The US Institute of Medicine (IOM) convened the Committee on the Learning Health Care System in America to explore these grand challenges. The committee’s report Best Care at Lower Cost issued call-to-action that focuses on “the rising complexity of modern health care, unsustainable cost increases, and outcomes below the system’s potential.”
IOM proposes the following characteristics of a continuously learning health care system:
§ Digital capture of the care experience for real-time generation and application of knowledge
§ Focus on patient needs and perspectives
§ Promoting the inclusion of patients, families and other caregivers as vital members of the continuously learning care team
§ Continuously aligning incentives for high-value care.
§ Systematically monitoring the safety, quality, processes, prices, costs and outcomes of care
§ Transparency of monitoring data to clinicians, patients and their families
§ Leadership-instilled oversight of learning, teamwork, collaboration and adaptability in support of continuous learning
§ Constantly refines complex care operations and processes through ongoing team training and skill building
Seeking to compress the average cycle time of 17 years from discovery in basic research to impact clinical care, Deloitte offers a four-phased perpetual model of learning health care system. Running on data, this rolling-wheel model consists of Clinical Research, Clinical Care, Health Outcomes and Surveillance, and Basic Research.
Practice guidelines that are generated via large research studies and applied to diverse patient populations attempt to standardize the clinical care process. Yet new knowledge and actionable reasoning that follow new discoveries do not always build up in a linear fashion. Dr. Nulan makes reference to Dr. Epstein’s concern that notwithstanding ongoing progress with medical knowledge and clinician’s ability to apply it for patient care, “too often the espoused remedies of one era [were proven] to be of limited value or frankly harmful in the next.” 
Data driven medicine is exemplified by Dr. Eugene Stead’s effort to change medical practice of patients with heart diseases from relying on anecdotal observations to evidence-based medicine; and the practice-based National Cardiovascular Data Registry which supports the outcomes-based quality improvement program of the American College of Cardiology.
Enabling technologies for medicine as a learning system offer evolving as well as disruptive capabilities. While a comprehensive account of these technologies is beyond the scope of this book, select topics are further discussed in the technology chapter. Examples of these technologies include electronic medical record, electronic health record, hospital information system, computerized physician order entry, personal health record, health information exchange, gene sequencing, gene browser, ontology, semantic web, health 2.0, social media, machine learning, analytics, big data, cloud-based computing, parallel computing, high performance computing, human factor engineering, data visualization, image analysis, signal processing, data storage, search engines, national language processing, learning systems, disease modeling, predictive analytics, security, etc.
‘The Future of Health Technology Over a 30 Year Span’ offers a 360-style infographic that further illuminates the topic.
And so, an open-source semantic network of personalized medicine content established in partnership between the public and private sector could help close the informatics gap in the field.
Security and confidentiality of health information in general and recent increase of regulatory burden on Protected Health Information (PHI) in particular require the kind of comprehensive discussion that is beyond the scope of this book. Instead, in addition to indicating the tremendous importance of this subject we offer a couple of recommended readings.
The International Standards Organization (ISO) identifies threats and specifies best practice guidelines for controlling and managing health information security.
“By implementing this International Standard, healthcare organizations and other custodians of health information will be able to ensure a minimum requisite level of security that is appropriate to their organization's circumstances and that will maintain the confidentiality, integrity and availability of personal health information.”
The NIST provides guidelines on security and privacy in public cloud computing.