E.H. Shortliffe and m.s. blois Fig 1.9"The Radio long before observers were ents ggesting how doctors and patients could communicate RADIO s ae pril using advanced technologies. This 1924 example is from l924 the cover of a popular Over 200 Illustrations magazine and envisions video enhancements Edited by H GERNSBACK to radio( Source: Radio THE RADIO DOCTOR-Maybe! 5P140 IN THIS ISSUE J. A. Fleming, F.R.S. d s. Pyle THE 100% RADIO MAGAZINE ULATION LARGER THAN ANY OTHER RADIO PUBLICATION more training programs, expansion of those that ers regarding the role of specialized multi- already exist, plus support for junior faculty in disciplinary expertise in successful cl health science schools who may wish to pursue systems implementation. The health care system additional training in this area provides some of the most complex organizational structures in society(Begun and Zimmerman 1.2.4.2 Organizational and 2003), and it is simplistic to assume that off-the Management Change shelf products will be smoothly introduced into Second, as implied above, there needs to be aa new institution without major analysis, rede greater understanding among health care lead- sign, and cooperative joint-development efforts Underinvestment and a failure to understand the bA directory of some existing training programs requirements for process reengineering as part availableathttp://www.amia.org/education/programs-and.ofsoftwareimplementationaswellasproblems courses(Accessed 3/3/2013) with technical leadership and planning, account
16 more training programs, 6 expansion of those that already exist, plus support for junior faculty in health science schools who may wish to pursue additional training in this area. 1.2.4.2 Organizational and Management Change Second, as implied above, there needs to be a greater understanding among health care lead- 6 A directory of some existing training programs is available at http://www.amia.org/education/programs-andcourses (Accessed 3/3/2013). ers regarding the role of specialized multidisciplinary expertise in successful clinical systems implementation. The health care system provides some of the most complex organizational structures in society (Begun and Zimmerman 2003 ), and it is simplistic to assume that off-theshelf products will be smoothly introduced into a new institution without major analysis, redesign, and cooperative joint-development efforts. Underinvestment and a failure to understand the requirements for process reengineering as part of software implementation, as well as problems with technical leadership and planning, account Fig. 1.9 “The Radio Doctor”: long before television was invented, creative observers were suggesting how doctors and patients could communicate using advanced technologies. This 1924 example is from the cover of a popular magazine and envisions video enhancements to radio (Source: “Radio News” 1924) E.H. Shortliffe and M.S. Blois
Biomedical Informatics: The Science and the Pragmatics 17 for many of the frustrating experiences that 1.3 The US Government Steps In health care organizations report in their efforts to use computers more effectively in support of During the early decades of the evolution of clinical patient care and provider productivity information systems for use in hospitals, patient The notion of a learning health care system care, and public health, the major role of govern described previously is meant to motivate your ment was in supporting the research enterprise as enthusiasm for what lies ahead and to suggest new methods were developed, tested, and formally the topics that need to be addressed in a book evaluated. The topic was seldom mentioned by the such as this one. Essentially all of the follow- nations leaders, however, even during the 1990s ing chapters touch on some aspect of the vision when the White House was viewed as being espe- of integrated systems that extend beyond single cially tech savvy. It was accordingly remarkable institutions. Before embarking on these topics, when, in the President 's State of the Union address however, we must emphasize two points. First, in 2004(and in each of the following years of his he cyclical creation of new knowledge in a administration), President Bush called for univer- learning health care system will become reality sal implementation of electronic health records only if individual hospitals, academic medical within 10 years. The Secretary of Health and centers, and national coordinating bodies work Human Services, Tommy Thompson, was simi together to provide the standards, infrastructure, larly supportive and, in May 2004, created an entity and resources that are necessary. No individual intended to support the expansion of the use of system developer, vendor, or administrator can EHRs--the Office of the National Coordinator mandate the standards for connectivity, data for Health Information Technology (initially pooling, and data sharing implied by a learn- referred to by the full acronym ONCHiT, but later ing health care system. A national initiative shortened simply to ONC) of cooperative planning and implementation There was limited budget for ONC, although for computing and communications resources the organization served as a convening body for within and among institutions and clinics is EHR-related planning efforts and the National required before practitioners will have routine Health Information Infrastructure(see Chaps access to the information that they need(see 12, 13, and 27). The topic of EHRs subsequently Chap. 13). A recent federal incentive program became a talking point for both major candi- for EHR implementation is a first step in this dates during the Presidential election in 2008 direction(see Sect. 1.3). The criteria that are with strong bipartisan support. However, it was required for successful EHR implementation the American Recovery and Reinvestment Act are sensitive to the need for data integration, (ARRA) in early 2009, also known as the eco- public-health support, and a learning health nomic "Stimulus Bill", that first provided major care system. funding to provide fiscal incentives for health Second, although our presentation systems, hospitals, and providers to implement learning health care notion has focused EHRs in their practices. Such payments were clinician's view of integrated information access, made available, however, only when eligible orga other workers in the field have similar needs that nizations or individual practitioners implemented can be addressed in similar ways. The academic EHRs that were"certified"as meeting minimal research community has already developed and standards and when they could document that made use of much of the technology that needs to they were making"meaningful use"of those sys- be coalesced if the clinical user is to have similar tems You will see references to such certification access to data and information. There is also the and meaningful use criteria in many chapters patients view, which must be considered in the this volume. There is also a discussion of HIT notion of patient-centered health care that is now policy and the federal government in Chap. 27. broadly accepted and encouraged(Ozkaynak Although the process of EHR implementation is etal.2013) still ongoing at present, the trend is clear: because
17 for many of the frustrating experiences that health care organizations report in their efforts to use computers more effectively in support of patient care and provider productivity. The notion of a learning health care system described previously is meant to motivate your enthusiasm for what lies ahead and to suggest the topics that need to be addressed in a book such as this one. Essentially all of the following chapters touch on some aspect of the vision of integrated systems that extend beyond single institutions. Before embarking on these topics, however, we must emphasize two points. First, the cyclical creation of new knowledge in a learning health care system will become reality only if individual hospitals, academic medical centers, and national coordinating bodies work together to provide the standards, infrastructure, and resources that are necessary. No individual system developer, vendor, or administrator can mandate the standards for connectivity, data pooling, and data sharing implied by a learning health care system. A national initiative of cooperative planning and implementation for computing and communications resources within and among institutions and clinics is required before practitioners will have routine access to the information that they need (see Chap. 13). A recent federal incentive program for EHR implementation is a fi rst step in this direction (see Sect. 1.3 ). The criteria that are required for successful EHR implementation are sensitive to the need for data integration, public- health support, and a learning health care system. Second, although our presentation of the learning health care notion has focused on the clinician’s view of integrated information access, other workers in the fi eld have similar needs that can be addressed in similar ways. The academic research community has already developed and made use of much of the technology that needs to be coalesced if the clinical user is to have similar access to data and information. There is also the patient’s view, which must be considered in the notion of patient-centered health care that is now broadly accepted and encouraged ( Ozkaynak et al. 2013 ). 1.3 The US Government Steps In During the early decades of the evolution of clinical information systems for use in hospitals, patient care, and public health, the major role of government was in supporting the research enterprise as new methods were developed, tested, and formally evaluated. The topic was seldom mentioned by the nation’s leaders, however, even during the 1990s when the White House was viewed as being especially tech savvy. It was accordingly remarkable when, in the President’s State of the Union address in 2004 (and in each of the following years of his administration), President Bush called for universal implementation of electronic health records within 10 years. The Secretary of Health and Human Services, Tommy Thompson, was similarly supportive and, in May 2004, created an entity intended to support the expansion of the use of EHRs—the Offi ce of the National Coordinator for Health Information Technology (initially referred to by the full acronym ONCHIT, but later shortened simply to ONC). There was limited budget for ONC, although the organization served as a convening body for EHR-related planning efforts and the National Health Information Infrastructure (see Chaps. 12, 13, and 27). The topic of EHRs subsequently became a talking point for both major candidates during the Presidential election in 2008, with strong bipartisan support. However, it was the American Recovery and Reinvestment Act (ARRA) in early 2009, also known as the economic “Stimulus Bill”, that fi rst provided major funding to provide fi scal incentives for health systems, hospitals, and providers to implement EHRs in their practices. Such payments were made available, however, only when eligible organizations or individual practitioners implemented EHRs that were “certifi ed” as meeting minimal standards and when they could document that they were making “meaningful use” of those systems. You will see references to such certifi cation and meaningful use criteria in many chapters in this volume. There is also a discussion of HIT policy and the federal government in Chap. 27. Although the process of EHR implementation is still ongoing at present, the trend is clear: because 1 Biomedical Informatics: The Science and the Pragmatics
E.H. Shortliffe and m.s. blois Health IT jobs Healthcare jobs 88883=E∈88=5品9 ITECH AC ary 2009 50 OOOO O 与喜房皇与皇启皇 Fig. 1.10 Percent change in online health IT job o.2,May2012[http://www.healthit.gov/sites/ per month, relative to health care jobs and all jobs les/pdf/05 12_ONCDataBrief2_JobPostings ized to February 2009 when ARRA passed ( Source analysis of data from O'Reilly Job Data Mart, ONC Data of the federal stimulus package, large numbers of has led to the development of many of the func- hospitals, systems, and practitioners are invest- tionalities that need to be brought together in the ing in EHRs and incorporating them into their integrated bio medical-computing environment practices. Furthermore, the demand for work- of the future. The remainder of this chapter deals ers skilled in health information technology has with biomedical informatics as a field and with grown much more rapidly than has the general biomedical and health information as a subject of job market, even within health care(Fig. 1.10). study. It provides additional background needed It is a remarkable example of how government to understand many of the subsequent chapters in policy and investment can stimulate major transi- this book tions in systems such as health care, where many Reference to the use of computers in bio- observers had previously felt that progress had medicine evokes different images depending been unacceptably slow(Shortliffe 2005). on the nature of ones involvement in the field To a hospital administrator, it might suggest the maintenance of clinical-care records using com 1.4 Defining Biomedical puters; to a decision scientist, it might mean the Informatics and related assistance by computers in disease diagnosis; to Disciplines a basic scientist, it might mean the use puters for maintaining, retrieving, and With the previous sections of this chapter as gene-sequencing information. Many background, let us now consider the scientific immediately think of office-practice tools for discipline that is the subject of this volume and tasks such as patient billing or appointment
18 of the federal stimulus package, large numbers of hospitals, systems, and practitioners are investing in EHRs and incorporating them into their practices. Furthermore, the demand for workers skilled in health information technology has grown much more rapidly than has the general job market, even within health care (Fig. 1.10 ). It is a remarkable example of how government policy and investment can stimulate major transitions in systems such as health care, where many observers had previously felt that progress had been unacceptably slow (Shortliffe 2005 ). 1.4 Defi ning Biomedical Informatics and Related Disciplines With the previous sections of this chapter as background, let us now consider the scientifi c discipline that is the subject of this volume and has led to the development of many of the functionalities that need to be brought together in the integrated bio medical-computing environment of the future. The remainder of this chapter deals with biomedical informatics as a fi eld and with biomedical and health information as a subject of study. It provides additional background needed to understand many of the subsequent chapters in this book. Reference to the use of computers in biomedicine evokes different images depending on the nature of one’s involvement in the fi eld. To a hospital administrator, it might suggest the maintenance of clinical-care records using computers; to a decision scientist, it might mean the assistance by computers in disease diagnosis; to a basic scientist, it might mean the use of computers for maintaining, retrieving, and analyzing gene-sequencing information. Many physicians immediately think of offi ce-practice tools for tasks such as patient billing or appointment 250 Health IT jobs HITECH Act February 2009 Healthcare jobs All jobs 199 57 52 200 150 100 50 0 –50 Percent change in Health IT job Positings per Month (normalized to Feb 2009) Jan-07 Mar-07 May-07 Jul-07 Sep-07 Nov-07 Jan-08 Mar-08 May-08 Jul-08 Sep-08 Nov-08 Jan-09 Mar-09 May-09 Jul-09 Sep-09 Nov-09 Jan-10 Mar-10 May-10 Jul-10 Sep-10 Nov-10 Jan-11 Mar-11 May-11 Jul-11 Sep-11 Nov-11 Jan-12 Fig. 1.10 Percent change in online health IT job postings per month, relative to health care jobs and all jobs: normalized to February 2009 when ARRA passed (Source: ONC analysis of data from O’Reilly Job Data Mart, ONC Data Brief, No. 2, May 2012 [http://www.healthit.gov/sites/ default/fi les/pdf/0512_ONCDataBrief2_JobPostings.pdf (Accessed 4/10/13)] E.H. Shortliffe and M.S. Blois
1 Biomedical Informatics: The Science and the Pragmatics scheduling. Nurses often think of computer-based given a sense of timeliness (if not urgency) by tools for charting the care that they deliver, or the simple existence of the computer. The cogni decision-support tools that assist in applying the tive activities of clinicians in practice probably most current patient-care guidelines. The field have received more attention over the past three includes study of all these activities and a great decades than in all previous history(see Chap many others too. More importantly, it includes 4). Again, the existence of the computer and the the consideration of various external factors that possibilities of its extending a clinicians cogni affect the biomedical setting. Unless you keep tive powers have motivated many of these stud- in mind these surrounding factors, it may be dif- ies. To develop computer-based tools to assist ficult to understand how biomedical computing with decisions, we must understand more clearly can help us to tie together the diverse aspects of such human processes as diagnosis, therapy health care and its delivery planning, decision making, and problem solving To achieve a unified perspective, we might in medicine. We must also understand how per consider four related topics: (1)the concept of sonal and cultural beliefs affect the way in which biomedical information(why it is important in information is interpreted and decisions are ulti biological research and clinical practice and why mately made we might want to use computers to process it); (2) the structural features of medicine, including all those subtopics to which computers might be 1.4.1 Terminology applied; (3) the importance of evidence-based knowledge of biomedical and health topics, Since the 1960s, by which time a growing number including its derivation and proper management of individuals doing serious biomedical research and use; and (4) the applications of computers or clinical practice had access to some kind of and communication methods in biomedicine and computer system, people have been uncertain he scientific issues that underlie such efforts. what name they should use for the biomedical We mention the first two topics briefly in this application of computer science concepts. The and the next chapter, and we provide references name computer science was itself new in 1960 the Suggested Readings section for those stu- and was only vaguely defined. Even today, the dents who wish to learn more. The third topic, term computer science is used more as a matter knowledge to support effective decision making of convention than as an explanation of the field in support of human health, is intrinsic to this scientific content book and occurs in various forms in essentially In the 1970s we began to use the phrase med- every chapter. The fourth topic, however, is the ical computer science to refer to the subdivision principal subject of this book. of computer science that applies the methods of Computers have captured the imagination the larger field to medical topics. As you will (and attention) of our society. Todays younger see, however, medicine has provided a rich area individuals have always lived in a world in which for computer science research, and several basic computers are ubiquitous and useful. Because computing insights and methodologies have the computer as a machine is exciting, people been derived from applied medical-computing may pay a disproportionate amount of atten- research. tion to it as such--at the expense of considering The term information science, which is occa what the computer can do given the numbers, sionally used in conjunction with computer sci- concepts, ideas, and cognitive underpinnings of ence, originated in the field of library science and fields such as medicine, health, and biomedical is used to refer, somewhat generally, to the broad research Computer scientists, philosophers, psy- range of issues related to the management of both chologists, and other scholars increasingly con- paper-based and electronically stored informa- sider such matters as the nature of information tion. Much of what information science origi and knowledge and how human beings process nally set out to be is now drawing evolvin such concepts. These investigations have been interest under the name cognitive science
19 scheduling. Nurses often think of computer- based tools for charting the care that they deliver, or decision-support tools that assist in applying the most current patient-care guidelines. The fi eld includes study of all these activities and a great many others too. More importantly, it includes the consideration of various external factors that affect the biomedical setting. Unless you keep in mind these surrounding factors, it may be dif- fi cult to understand how biomedical computing can help us to tie together the diverse aspects of health care and its delivery. To achieve a unifi ed perspective, we might consider four related topics: (1) the concept of biomedical information (why it is important in biological research and clinical practice and why we might want to use computers to process it); (2) the structural features of medicine, including all those subtopics to which computers might be applied; (3) the importance of evidence-based knowledge of biomedical and health topics, including its derivation and proper management and use; and (4) the applications of computers and communication methods in biomedicine and the scientifi c issues that underlie such efforts. We mention the fi rst two topics briefl y in this and the next chapter, and we provide references in the Suggested Readings section for those students who wish to learn more. The third topic, knowledge to support effective decision making in support of human health, is intrinsic to this book and occurs in various forms in essentially every chapter. The fourth topic, however, is the principal subject of this book. Computers have captured the imagination (and attention) of our society. Today’s younger individuals have always lived in a world in which computers are ubiquitous and useful. Because the computer as a machine is exciting, people may pay a disproportionate amount of attention to it as such—at the expense of considering what the computer can do given the numbers, concepts, ideas, and cognitive underpinnings of fi elds such as medicine, health, and biomedical research. Computer scientists, philosophers, psychologists, and other scholars increasingly consider such matters as the nature of information and knowledge and how human beings process such concepts. These investigations have been given a sense of timeliness (if not urgency) by the simple existence of the computer. The cognitive activities of clinicians in practice probably have received more attention over the past three decades than in all previous history (see Chap. 4). Again, the existence of the computer and the possibilities of its extending a clinician’s cognitive powers have motivated many of these studies. To develop computer-based tools to assist with decisions, we must understand more clearly such human processes as diagnosis, therapy planning, decision making, and problem solving in medicine. We must also understand how personal and cultural beliefs affect the way in which information is interpreted and decisions are ultimately made. 1.4.1 Terminology Since the 1960s, by which time a growing number of individuals doing serious biomedical research or clinical practice had access to some kind of computer system, people have been uncertain what name they should use for the biomedical application of computer science concepts. The name computer science was itself new in 1960 and was only vaguely defi ned. Even today, the term computer science is used more as a matter of convention than as an explanation of the fi eld’s scientifi c content. In the 1970s we began to use the phrase medical computer science to refer to the subdivision of computer science that applies the methods of the larger fi eld to medical topics. As you will see, however, medicine has provided a rich area for computer science research, and several basic computing insights and methodologies have been derived from applied medical-computing research. The term information science , which is occasionally used in conjunction with computer science, originated in the fi eld of library science and is used to refer, somewhat generally, to the broad range of issues related to the management of both paper-based and electronically stored information. Much of what information science originally set out to be is now drawing evolving interest under the name cognitive science . 1 Biomedical Informatics: The Science and the Pragmatics
E.H. Shortliffe and m.s. blois Information theory, in contrast, was first life-science professionals. Thus, the term health developed by scientists concerned about the informatics, or health care informatics, gained physics of communication; it has evolved into some popularity, even though it has the disadvan what may be viewed as a branch of mathematics. tage of tending to exclude applications to bio- The results scientists have obtained with infor- medical research( Chaps. 24 and 25)and, as we mation theory have illuminated many processes will argue shortly, it tends to focus the field's in communications technology, but they have had name on application domains(clinical care, pub- little effect on our understanding of human infor- lic health, and prevention) rather than the basic mation processing. discipline and its broad range of applicability The terms biomedical computing or biocom- Applications of informatics methods in biol putation have been used for a number of years. ogy and genetics exploded during the 1990s due hey are nondescriptive and neutral, imply- to the human genome projectand the growing ing only that computers are employed for some recognition that modern life-science research purpose in biology or medicine. They are often was no longer possible without computational associated with bioengineering applications support and analysis(see Chaps. 24 and 25) of computers, however, in which the devices By the late 1990s, the use of informatics meth- are viewed more as tools for a bioengineering ods in such work had become widely known as application than as a primary focus of research. bioinformatics and the director of the National In the 1970s, inspired by the French term for Institutes of Health(NIH)appointed an advisory computer science(informatique ), the English- group called the Working Group on Biomedical speaking community began to use the term medi- Computing. In June 1999, the group provided a cal informatics. Those in the field were attracted report recommending that the nih undertake by the words emphasis on information, which an initiative called the Biomedical Information they saw as more central to the field than the Science and Technology Initiative (BISTi) computer itself, and it gained momentum as a With the subsequent creation of another NIH term for the discipline, especially in Europe, dur- organization called the Bioinformatics Working ing the 1980s. The term is broader than medical Group, the visibility of informatics applications computing (it includes such topics as medical in biology was greatly enhanced. Today bioinfor- statistics, record keeping, and the study of the matics is a major area of activity at the Niland nature of medical information itself) and deem- in many universities and biotechnology compa- phasizes the computer while focusing instead on nies around the world. The explosive growth of the nature of the field to which computations are this field, however, has added to the confusion pplied. Because the term informatics became regarding the naming conventions we have widely accepted in the United States only in the discussing. In addition, the relationship between late 1980s, medical information science was medical informatics and bioinformatics became also used earlier in North America: this term. unclear. As a result in an effort to be more incl however, may be confused with library science, sive and to embrace the biological applications and it does not capture the broader implications with which many medical informatics groups of the European term. As a result, the name medi- had already been involved, the name medical al informatics appeared by the late 1980s to informatics gradually gave way to biomedical have become the preferred term, even in the informatics (BMI). Several academic groups United States. Indeed, this is the name of the field have changed their names, and a major medical that we used in the first two editions of this text- informatics journal( Computers and Biomedical book (from 1990 to 2000), and it is still some- times used in professional, industrial, and academicsettingsHowever,manyobservershttp://www.ornl.gow/sci/techresources/human_genome/ expressedconcernthattheadjective"medical"is8avaIlableathttp://www.nih.gow/about/director/060399 too focused on physicians and fails to appreciate html(Accessed 4/8/2013) therelevanceofthisdisciplinetootherhealthandseehttp://www.bisti.nih.gov/.(accessed4/8/2013)
20 Information theory , in contrast, was fi rst developed by scientists concerned about the physics of communication; it has evolved into what may be viewed as a branch of mathematics. The results scientists have obtained with information theory have illuminated many processes in communications technology, but they have had little effect on our understanding of human information processing. The terms biomedical computing or biocomputation have been used for a number of years. They are nondescriptive and neutral, implying only that computers are employed for some purpose in biology or medicine. They are often associated with bioengineering applications of computers, however, in which the devices are viewed more as tools for a bioengineering application than as a primary focus of research. In the 1970s, inspired by the French term for computer science ( informatique ), the Englishspeaking community began to use the term medical informatics . Those in the fi eld were attracted by the word’s emphasis on information , which they saw as more central to the fi eld than the computer itself, and it gained momentum as a term for the discipline, especially in Europe, during the 1980s. The term is broader than medical computing (it includes such topics as medical statistics, record keeping, and the study of the nature of medical information itself) and deemphasizes the computer while focusing instead on the nature of the fi eld to which computations are applied. Because the term informatics became widely accepted in the United States only in the late 1980s, medical information science was also used earlier in North America; this term, however, may be confused with library science, and it does not capture the broader implications of the European term. As a result, the name medical informatics appeared by the late 1980s to have become the preferred term, even in the United States. Indeed, this is the name of the fi eld that we used in the fi rst two editions of this textbook (from 1990 to 2000), and it is still sometimes used in professional, industrial, and academic settings. However, many observers expressed concern that the adjective “medical” is too focused on physicians and fails to appreciate the relevance of this discipline to other health and life-science professionals. Thus, the term health informatics , or health care informatics, gained some popularity, even though it has the disadvantage of tending to exclude applications to biomedical research (Chaps. 24 and 25) and, as we will argue shortly, it tends to focus the fi eld’s name on application domains (clinical care, public health, and prevention) rather than the basic discipline and its broad range of applicability. Applications of informatics methods in biology and genetics exploded during the 1990s due to the human genome project 7 and the growing recognition that modern life-science research was no longer possible without computational support and analysis (see Chaps. 24 and 25). By the late 1990s, the use of informatics methods in such work had become widely known as bioinformatics and the director of the National Institutes of Health (NIH) appointed an advisory group called the Working Group on Biomedical Computing. In June 1999, the group provided a report 8 recommending that the NIH undertake an initiative called the Biomedical Information Science and Technology Initiative ( BISTI ). With the subsequent creation of another NIH organization called the Bioinformatics Working Group, the visibility of informatics applications in biology was greatly enhanced. Today bioinformatics is a major area of activity at the NIH 9 and in many universities and biotechnology companies around the world. The explosive growth of this fi eld, however, has added to the confusion regarding the naming conventions we have been discussing. In addition, the relationship between medical informatics and bioinformatics became unclear. As a result, in an effort to be more inclusive and to embrace the biological applications with which many medical informatics groups had already been involved, the name medical informatics gradually gave way to biomedical informatics (BMI). Several academic groups have changed their names, and a major medical informatics journal ( Computers and Biomedical 7 http://www.ornl.gov/sci/techresources/Human_Genome/ home.shtml (Accessed 4/8/2013). 8 Available at http://www.nih.gov/about/director/060399. html (Accessed 4/8/2013). 9 See http://www.bisti.nih.gov/ . (Accessed 4/8/2013). E.H. Shortliffe and M.S. Blois