- Email: [email protected]
Overview of the ISCE ECG “genome project”
Journal of Electrocardiology Vol. 36 Supplement 2003
Overview of the ISCE ECG “Genome Project”
Paul Kligfield, MD
Abstract: The International Society for Computerized Electrocardiography (ISCE) “genome” project was begun at the end of 2000 to explore mechanisms for development of a cross-platform electrocardiogram (ECG) database. Ultimate feasibility of this project is based on established interactive cooperation of clinical investigators, engineers, and industry within the ISCE framework. The US Food and Drug Administration (FDA) mandate for centralized access to digitized ECG waveforms used in clinical trials provides a complementary stimulus to technologies that facilitate ECG database development. The constituency of ISCE is interested in acquisition and analysis of data from both standard 12-lead (resting) ECG and ambulatory (monitoring) ECG. Support for project goals from industry, at the Trustee as well as at the engineering level, has led to initial focus on the resting ECG. A one-year pilot project has been proposed to establish and implement software methodology for transmission, storage, and integrated reanalysis of digitized ECG waveforms provided by several major manufacturers. Beyond data acquisition, storage, and analysis, a number of critical issues are associated with database development. These include definition of clinically relevant “gold” standards, acquisition and validation of non-ECG data, protection of patient privacy, control and ownership of data, and accessibility and use of the database. However, implementation of the pilot project is a necessary first step, since all issues become moot without technical cooperation for shared formatting and analysis. Key words: ECG, database, population studies.
A century after Willem Einthoven first “sequenced” the electrocardiogram (ECG), the ISCE “genome” project was begun to explore mechanisms for development of a database that would be useful to our academic and industry constituents. The initial charge from the ISCE Board was direc-
tionally nonspecific and open-ended. The ISCE constituency is interested in improvement of the clinical value of the ECG, development of new ECG criteria, population studies, testing and validation of new devices, and marketing equipment. With respect to surface recording of the electrocardiogram, members are interested in both standard 12-lead (resting) ECG data and, separately, ambulatory (monitoring) ECG data. While most standard ECG manufacturers are large companies with additional interest in monitoring devices, some of the monitoring companies are not involved with standard ECG. Although there is considerable overlap in engineering between these methodologies, their
From the Division of Cardiology, Department of Medicine, Weill Medical College of Cornell University, New York, NY. Reprint requests: Paul Kligfield, MD, Division of Cardiology, Department of Medicine, Weill Medical College of Cornell University, New York, NY; e-mail: [email protected]. © 2003 Elsevier Inc. All rights reserved. 0022-0736/03/360S-0041$30.00/0 doi:10.1016/j.jelectrocard.2003.09.042
163
164 Journal of Electrocardiology Vol. 36 Supplement 2003 endpoint analyses and database needs differ markedly. Of course, it was recognized at the outset that a number of established ECG databases and database tools are in existence (1,2). Many of these projects have involved members of ISCE, but none have been primarily sponsored or facilitated by ISCE. By way of example, the ECG database registry of The National Heart, Lung, and Blood Institute (NHLBI, www.nhlbi.nih.gov/ecgdata) provides links to four resources, including digitized ECGs from prior examination cycles of the Framingham Heart Study, 12 lead records from healthy adult Chinese men and women, and 2 sources of 12 lead records from catheterized patients with and without myocardial infarction. Links to additional resources are provided by PhysioBank, a service of the National Center for Research Resources of the National Institutes of Health (NIH) (www.physionet.org/physiobank), including the AHA database of annotated analog ambulatory ECG recordings, the Ann Arbor Electrogram Libraries of annotated intracardiac electrograms, the Common Standards for Electrocardiography (CSE) database of 12- and 15-lead ECGs, and two collections of multiple physiologic signals from the critical care monitoring environment. Standard multilead ECGs represent “cross-sectional” individual data in the sense that recording time is brief, but the database may be “longitudinal” when serial studies within individuals are included. Progress has been made toward standardizing formats for these purposes during the past decade. Based on collaboration with a number of manufacturers, a standard communications protocol for interchange, encoding, and storage of digital ECG data was proposed in 1992 (3). As observed by Zywietz in a subsequent report (4), the main objective of integration of information requires interconnectivity and interoperability of devices, which can only be achieved by the development of common communication standards. Recent interest of the US Food and Drug Administration (FDA) in collection and review of digital ECG waveforms has also focused industry attention on XML technology (5). Monitoring type databases examine longer term recordings of annotated rhythm and waveform findings such as ST-segment deviation (6). A description of these sources, including strengths and weaknesses, is available in an AAMI document (7). Although demographics of these data sources have been limited, progress in acquisition of pediatric intensive care monitoring data has been recently reported (8), while progress toward establishment
of a standard output file format for ambulatory ECG data has been made by ISHNE (9,10). Pilot discussions regarding database needs were coordinated by ISCE Secretary Paul Kligfield and were held by e-mail iteration in early 2001 with a group of ISCE member volunteers representing a variety of clinical, engineering, and manufacturing interests. Initial participants included Rosalie Dunn, Jouni Erkkila, then ISCE President Michael Laks, Don Lin, Simon Meij, Cadathur Rajagopalan, Shankara Reddy, Ron Selvester, Robert Warner, and Sophia Zhou. Additional discussions and formation of working groups were held at the 2001 ISCE meeting at Hutchinson Island. These discussions revealed that exploration of a new “cross-platform” ECG database under ISCE auspices is desirable and feasible. At the same time, it was recognized that a number of clinical, engineering, administrative, and business challenges require solution before a useful and dynamic database could be implemented. Discussions among the ISCE volunteer group indicated that despite progress regarding standards, proposed uniform ECG data formats have not achieved industry consensus and user acceptance. Limitations of available resources were noted to include restricted or costly access, proprietary signal management, occasionally small numbers of patients, variability of the nature and adequacy of “gold standard” ascertainment, and lack of demographic range. Perhaps most important, available databases have often been “nondynamic,” in the sense that growth over time has been difficult. Established databases have not prominently encouraged ongoing, multicenter contribution of digitized ECG information that was obtained with different types of standard equipment. Accordingly, it was generally agreed that a “cross-platform” 12-lead ECG database was desirable, meaning that it would be optimal from a clinical perspective and possible from an engineering perspective to incorporate data obtained with different ECG technologies into a single resource. This “cross-platform” database should be capable of growth and facilitate reanalysis by newly applied diagnostic algorithms. Similar issues and limitations were noted to affect the development of a “cross-platform” database of monitoring data. Initial issues that emerged with respect the standard ECG were summarized by Sophia Zhou. She noted that engineers present at the initial meeting from Philips Medical Systems, GE Medical Systems, and Mortara Instrument were strongly interested in supporting the “cross-platform” 12-lead database concept. However, it was clear from the very outset that cooperation with the “cross-platform” process
The ISCE “Genome Project” •
among manufacturers was a business decision as well as an engineering challenge. XML file format, an increasingly used markup language for documents containing structured information, was recommended as a timely choice for ECG data exchange and storage. It was suggested that ECGs collected from different cardiographs may be converted into XML format, and that ECGs in CSE or ICE format could be submitted to an ISCE 12-lead ECG processing center for conversion into XML format (for more on XML, see www.xml.com/pub/ a/98/10/guide0.html). Issues relating to “long-term recording” databases were summarized by Cadathur Rajagopalan. Problems for resolution include selection of a format for the monitoring database and consensus regarding beat and rhythm annotation and ST-segment measurement. It also was emphasized that perception and value of a monitoring database will depend on the user. Users within the medical device industry tend to look at databases as means for beat-to-beat testing and validation of software algorithms that often are proprietary. In contrast, users within the academic world tend to visualize these databases as opportunities for the development of new research ideas. While only several manufacturers participated in these initial discussions, representatives from other ECG manufacturers present at the meeting were invited to participate in the database development process. Beyond data acquisition, storage, and analysis, a number of critical issues are associated with database development. These include definition of clinically relevant “gold” standards, acquisition and validation of non-ECG data, protection of patient privacy, control and ownership of data, and accessibility and use of the database. Will data need to be generated de novo for our purposes? Will it be possible to import data generated by prior and ongoing studies with validated demographics and varying endpoints? Will federal and pharmaceutical industry sponsors of relevant projects share information with such a project? Will the scope of the project mandate federal support? Who will control access to information provided by many sources? These exceptionally complex issues are probably solvable, but all become moot without technical cooperation for shared formatting and analysis of ECG data. Inertia and limitations notwithstanding, the ultimate feasibility of a cooperative database project is rooted in the established productive interaction of clinical investigators, engineers, and industry within the ISCE framework. The FDA mandate for
Paul Kligfield 165
centralized access to digitized ECG waveforms used in clinical drug trials provides a complementary stimulus to technologies that facilitate ECG database development. Support for project goals from industry, at the Trustee as well as at the engineering level, has led to initial focus on the resting 12-lead ECG. A 1-year pilot project has been proposed to establish and implement software methodology for transmission, storage, and integrated reanalysis of digitized ECG waveforms provided by several major manufacturers. Discussions during the current ISCE meeting will focus on moving forward in this direction.
References 1. Norman JE, Bailey JJ, Berson AS, et al: NHLBI workshop on the utilization of ECG databases: preservation and use of existing ECG databases and development of future resources. J Electrocardiol 31:83, 1998 2. Goldberger AL, Amaral LA, Glass L, et al: PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals. Circulation 101:E215, 2000 3. Willems JL, Zywietz C, Rubel P, et al: A standard communications protocol for computerized electrocardiography. J Electrocardiol 24:173, 1992 (suppl) 4. Zywietz C: SCP-ECG and Vital Signs Information Representation–Two examples of successful transcontinental cooperation in medical informatics standardization. Int J Med Inf 48:195, 1998 5. Department of Health and Human Services, Food and Drug Administration. Docket No. 01N-0476. Electronic interchange standard for digital ECG and similar data; public meeting. Federal Register 2001;26 (No 206):53801-53802. 6. Jager F, Taddei A, Moody GB, et al: Long-term ST database: a reference for the development and evaluation of automated ischaemia detectors and for the study of the dynamics of myocardial ischaemia. Med Biol Eng Comput 41:172, 2003 7. AAMI. Acquisition and use of physiologic waveform databases for testing of medical devices. TIR 24:1999 8. Goldstein B, McNames J, McDonald BA, et al: Physiologic data acquisition system and database for the study of disease dynamics in the intensive care unit. Crit Care Med 31:433, 2003 9. Zareba W, Locati EH, Maison Blanche P, for the ISHNE Holter Standard Output File Format Task Force: The ISHNE Holter standard output file format: A step toward compatibility of Holter systems. Ann Noninv Electrocardiol 5:261, 1998 10. Badilini F, for the ISHNE Standard Output Format Task Force: The ISHNE Holter standard output file format. Ann Noninv Electrocardiol 5:263, 1998
Overview of the ISCE ECG “Genome Project”
Paul Kligfield, MD
Abstract: The International Society for Computerized Electrocardiography (ISCE) “genome” project was begun at the end of 2000 to explore mechanisms for development of a cross-platform electrocardiogram (ECG) database. Ultimate feasibility of this project is based on established interactive cooperation of clinical investigators, engineers, and industry within the ISCE framework. The US Food and Drug Administration (FDA) mandate for centralized access to digitized ECG waveforms used in clinical trials provides a complementary stimulus to technologies that facilitate ECG database development. The constituency of ISCE is interested in acquisition and analysis of data from both standard 12-lead (resting) ECG and ambulatory (monitoring) ECG. Support for project goals from industry, at the Trustee as well as at the engineering level, has led to initial focus on the resting ECG. A one-year pilot project has been proposed to establish and implement software methodology for transmission, storage, and integrated reanalysis of digitized ECG waveforms provided by several major manufacturers. Beyond data acquisition, storage, and analysis, a number of critical issues are associated with database development. These include definition of clinically relevant “gold” standards, acquisition and validation of non-ECG data, protection of patient privacy, control and ownership of data, and accessibility and use of the database. However, implementation of the pilot project is a necessary first step, since all issues become moot without technical cooperation for shared formatting and analysis. Key words: ECG, database, population studies.
A century after Willem Einthoven first “sequenced” the electrocardiogram (ECG), the ISCE “genome” project was begun to explore mechanisms for development of a database that would be useful to our academic and industry constituents. The initial charge from the ISCE Board was direc-
tionally nonspecific and open-ended. The ISCE constituency is interested in improvement of the clinical value of the ECG, development of new ECG criteria, population studies, testing and validation of new devices, and marketing equipment. With respect to surface recording of the electrocardiogram, members are interested in both standard 12-lead (resting) ECG data and, separately, ambulatory (monitoring) ECG data. While most standard ECG manufacturers are large companies with additional interest in monitoring devices, some of the monitoring companies are not involved with standard ECG. Although there is considerable overlap in engineering between these methodologies, their
From the Division of Cardiology, Department of Medicine, Weill Medical College of Cornell University, New York, NY. Reprint requests: Paul Kligfield, MD, Division of Cardiology, Department of Medicine, Weill Medical College of Cornell University, New York, NY; e-mail: [email protected]. © 2003 Elsevier Inc. All rights reserved. 0022-0736/03/360S-0041$30.00/0 doi:10.1016/j.jelectrocard.2003.09.042
163
164 Journal of Electrocardiology Vol. 36 Supplement 2003 endpoint analyses and database needs differ markedly. Of course, it was recognized at the outset that a number of established ECG databases and database tools are in existence (1,2). Many of these projects have involved members of ISCE, but none have been primarily sponsored or facilitated by ISCE. By way of example, the ECG database registry of The National Heart, Lung, and Blood Institute (NHLBI, www.nhlbi.nih.gov/ecgdata) provides links to four resources, including digitized ECGs from prior examination cycles of the Framingham Heart Study, 12 lead records from healthy adult Chinese men and women, and 2 sources of 12 lead records from catheterized patients with and without myocardial infarction. Links to additional resources are provided by PhysioBank, a service of the National Center for Research Resources of the National Institutes of Health (NIH) (www.physionet.org/physiobank), including the AHA database of annotated analog ambulatory ECG recordings, the Ann Arbor Electrogram Libraries of annotated intracardiac electrograms, the Common Standards for Electrocardiography (CSE) database of 12- and 15-lead ECGs, and two collections of multiple physiologic signals from the critical care monitoring environment. Standard multilead ECGs represent “cross-sectional” individual data in the sense that recording time is brief, but the database may be “longitudinal” when serial studies within individuals are included. Progress has been made toward standardizing formats for these purposes during the past decade. Based on collaboration with a number of manufacturers, a standard communications protocol for interchange, encoding, and storage of digital ECG data was proposed in 1992 (3). As observed by Zywietz in a subsequent report (4), the main objective of integration of information requires interconnectivity and interoperability of devices, which can only be achieved by the development of common communication standards. Recent interest of the US Food and Drug Administration (FDA) in collection and review of digital ECG waveforms has also focused industry attention on XML technology (5). Monitoring type databases examine longer term recordings of annotated rhythm and waveform findings such as ST-segment deviation (6). A description of these sources, including strengths and weaknesses, is available in an AAMI document (7). Although demographics of these data sources have been limited, progress in acquisition of pediatric intensive care monitoring data has been recently reported (8), while progress toward establishment
of a standard output file format for ambulatory ECG data has been made by ISHNE (9,10). Pilot discussions regarding database needs were coordinated by ISCE Secretary Paul Kligfield and were held by e-mail iteration in early 2001 with a group of ISCE member volunteers representing a variety of clinical, engineering, and manufacturing interests. Initial participants included Rosalie Dunn, Jouni Erkkila, then ISCE President Michael Laks, Don Lin, Simon Meij, Cadathur Rajagopalan, Shankara Reddy, Ron Selvester, Robert Warner, and Sophia Zhou. Additional discussions and formation of working groups were held at the 2001 ISCE meeting at Hutchinson Island. These discussions revealed that exploration of a new “cross-platform” ECG database under ISCE auspices is desirable and feasible. At the same time, it was recognized that a number of clinical, engineering, administrative, and business challenges require solution before a useful and dynamic database could be implemented. Discussions among the ISCE volunteer group indicated that despite progress regarding standards, proposed uniform ECG data formats have not achieved industry consensus and user acceptance. Limitations of available resources were noted to include restricted or costly access, proprietary signal management, occasionally small numbers of patients, variability of the nature and adequacy of “gold standard” ascertainment, and lack of demographic range. Perhaps most important, available databases have often been “nondynamic,” in the sense that growth over time has been difficult. Established databases have not prominently encouraged ongoing, multicenter contribution of digitized ECG information that was obtained with different types of standard equipment. Accordingly, it was generally agreed that a “cross-platform” 12-lead ECG database was desirable, meaning that it would be optimal from a clinical perspective and possible from an engineering perspective to incorporate data obtained with different ECG technologies into a single resource. This “cross-platform” database should be capable of growth and facilitate reanalysis by newly applied diagnostic algorithms. Similar issues and limitations were noted to affect the development of a “cross-platform” database of monitoring data. Initial issues that emerged with respect the standard ECG were summarized by Sophia Zhou. She noted that engineers present at the initial meeting from Philips Medical Systems, GE Medical Systems, and Mortara Instrument were strongly interested in supporting the “cross-platform” 12-lead database concept. However, it was clear from the very outset that cooperation with the “cross-platform” process
The ISCE “Genome Project” •
among manufacturers was a business decision as well as an engineering challenge. XML file format, an increasingly used markup language for documents containing structured information, was recommended as a timely choice for ECG data exchange and storage. It was suggested that ECGs collected from different cardiographs may be converted into XML format, and that ECGs in CSE or ICE format could be submitted to an ISCE 12-lead ECG processing center for conversion into XML format (for more on XML, see www.xml.com/pub/ a/98/10/guide0.html). Issues relating to “long-term recording” databases were summarized by Cadathur Rajagopalan. Problems for resolution include selection of a format for the monitoring database and consensus regarding beat and rhythm annotation and ST-segment measurement. It also was emphasized that perception and value of a monitoring database will depend on the user. Users within the medical device industry tend to look at databases as means for beat-to-beat testing and validation of software algorithms that often are proprietary. In contrast, users within the academic world tend to visualize these databases as opportunities for the development of new research ideas. While only several manufacturers participated in these initial discussions, representatives from other ECG manufacturers present at the meeting were invited to participate in the database development process. Beyond data acquisition, storage, and analysis, a number of critical issues are associated with database development. These include definition of clinically relevant “gold” standards, acquisition and validation of non-ECG data, protection of patient privacy, control and ownership of data, and accessibility and use of the database. Will data need to be generated de novo for our purposes? Will it be possible to import data generated by prior and ongoing studies with validated demographics and varying endpoints? Will federal and pharmaceutical industry sponsors of relevant projects share information with such a project? Will the scope of the project mandate federal support? Who will control access to information provided by many sources? These exceptionally complex issues are probably solvable, but all become moot without technical cooperation for shared formatting and analysis of ECG data. Inertia and limitations notwithstanding, the ultimate feasibility of a cooperative database project is rooted in the established productive interaction of clinical investigators, engineers, and industry within the ISCE framework. The FDA mandate for
Paul Kligfield 165
centralized access to digitized ECG waveforms used in clinical drug trials provides a complementary stimulus to technologies that facilitate ECG database development. Support for project goals from industry, at the Trustee as well as at the engineering level, has led to initial focus on the resting 12-lead ECG. A 1-year pilot project has been proposed to establish and implement software methodology for transmission, storage, and integrated reanalysis of digitized ECG waveforms provided by several major manufacturers. Discussions during the current ISCE meeting will focus on moving forward in this direction.
References 1. Norman JE, Bailey JJ, Berson AS, et al: NHLBI workshop on the utilization of ECG databases: preservation and use of existing ECG databases and development of future resources. J Electrocardiol 31:83, 1998 2. Goldberger AL, Amaral LA, Glass L, et al: PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals. Circulation 101:E215, 2000 3. Willems JL, Zywietz C, Rubel P, et al: A standard communications protocol for computerized electrocardiography. J Electrocardiol 24:173, 1992 (suppl) 4. Zywietz C: SCP-ECG and Vital Signs Information Representation–Two examples of successful transcontinental cooperation in medical informatics standardization. Int J Med Inf 48:195, 1998 5. Department of Health and Human Services, Food and Drug Administration. Docket No. 01N-0476. Electronic interchange standard for digital ECG and similar data; public meeting. Federal Register 2001;26 (No 206):53801-53802. 6. Jager F, Taddei A, Moody GB, et al: Long-term ST database: a reference for the development and evaluation of automated ischaemia detectors and for the study of the dynamics of myocardial ischaemia. Med Biol Eng Comput 41:172, 2003 7. AAMI. Acquisition and use of physiologic waveform databases for testing of medical devices. TIR 24:1999 8. Goldstein B, McNames J, McDonald BA, et al: Physiologic data acquisition system and database for the study of disease dynamics in the intensive care unit. Crit Care Med 31:433, 2003 9. Zareba W, Locati EH, Maison Blanche P, for the ISHNE Holter Standard Output File Format Task Force: The ISHNE Holter standard output file format: A step toward compatibility of Holter systems. Ann Noninv Electrocardiol 5:261, 1998 10. Badilini F, for the ISHNE Standard Output Format Task Force: The ISHNE Holter standard output file format. Ann Noninv Electrocardiol 5:263, 1998