Archives of Physical Medicine and Rehabilitation journal homepage: www.archives-pmr.org Archives of Physical Medicine and Rehabilitation 2015;96(3 Suppl 1):S8-15
SPECIAL COMMUNICATION
Critical Issues Using Brain-Computer Interfaces for Augmentative and Alternative Communication Katya Hill, PhD, CCC-SLP,a,b Thomas Kovacs, MA,a,b Sangeun Shin, MSa,b From aResearch Services, Pittsburgh VA HealthCare System, Pittsburgh, PA; and bAAC Performance, Testing and Teaching Lab, Communication Science and Disorders, School of Health and Rehabilitation Sciences, University of Pittsburgh, Pittsburgh, PA.
Abstract Brain-computer interfaces (BCIs) may potentially be of significant practical value to patients in advanced stages of amyotrophic lateral sclerosis and locked-in syndrome for whom conventional augmentative and alternative communication (AAC) systems, which require some measure of consistent voluntary muscle control, are not satisfactory options. However, BCIs have primarily been used for communication in laboratory research settings. This article discusses 4 critical issues that should be addressed as BCIs are translated out of laboratory settings to become fully functional BCI/AAC systems that may be implemented clinically. These issues include (1) identification of primary, secondary, and tertiary system features; (2) integrating BCI/AAC systems in the World Health Organization’s International Classification of Functioning, Disability and Health framework; (3) implementing language-based assessment and intervention; and (4) performance measurement. A clinical demonstration project is presented as an example of research beginning to address these critical issues. Archives of Physical Medicine and Rehabilitation 2015;96(3 Suppl 1):S8-15 ª 2015 by the American Congress of Rehabilitation Medicine
Augmentative and alternative communication (AAC) is a field of endeavor dedicated to providing effective and efficient interventions, strategies, and technology to help individuals whose natural speech is not functional to participate in their daily activities.1 The most effective AAC interventions supplement or replace speech production, or both, with strategies and technology that offer characteristics of a natural language; for example, the AAC system allows for spontaneous, novel utterance generation.2 Spontaneous novel utterance generation allows a person to formulate sentences and say anything at any time,3 and thus achieve interactive communication. Communication generated using AAC systems can be measured using traditional measures of language performance to gauge overall effectiveness in addition to measures of achieved functional outcomes. Additionally, AAC system features can be selected and
Presented to the National Institutes of Health, National Science Foundation, and other organizations (for a full list, see http://bcimeeting.org/2013/sponsors.html), June 3-7, 2013, Asilomar Conference Grounds, Pacific Grove, CA. Supported in part by the National Institutes of Health (grant no. R123-DE-01274401). The BCI study used as an example in this article was the VA CSP #567 project, funded by the VA Cooperative Studies Program, Department of Veterans Affairs Office of Research and Development. Specific values are reported by the Pittsburgh VAMC Laboratory. Disclosures: none.
manipulated to improve the communication performance of the AAC speaker or system. Brain-computer interfaces (BCIs) may potentially be of significant practical value to patients in advanced stages of amyotrophic lateral sclerosis (ALS) and locked-in syndrome for whom conventional AAC systems, which require some measure of consistent voluntary muscle control, are not satisfactory options. In these individuals, brain signals (eg, P300 potentials) might be good alternatives to channel as a noninvasive approach for accessing assistive technologies. However, measuring the effectiveness of BCIs to support communication becomes an expectation as BCIs mature into fully functional BCI/AAC systems.4 As translational research efforts move BCIs out of laboratory settings to be recommended by speech-language pathologists as viable BCI/AAC systems, quantitative language and communication metrics will become critical for making informed decisions about effectiveness and value. BCI research and development and clinical AAC practice benefit from the use of clearly defined or standardized measures for comparing performance between and within AAC speakers and systems. The purpose of this article is to introduce 4 critical issues that the BCI field should be accountable for addressing as BCI systems mature into BCI/AAC systems that may be implemented
0003-9993/14/$36 - see front matter ª 2015 by the American Congress of Rehabilitation Medicine http://dx.doi.org/10.1016/j.apmr.2014.01.034
Critical issues using brain-computer interfaces clinically. These issues include (1) identification of primary, secondary, and tertiary features of BCI/AAC systems; (2) integrating BCI/AAC systems in the World Health Organization’s International Classification of Functioning, Disability and Health (ICF) framework5; (3) implementing language-based assessment and intervention; and (4) performance measurement. A clinical demonstration project measuring the communication performance of BCI/AAC system speakers6 is presented as an example of research beginning to address these critical issues.
Critical issues to address Primary, secondary, and tertiary device features The first critical issue to address relates to identifying features of BCI/AAC systems. AAC interventions are compared and matched to an individual based on primary, secondary, and tertiary features. These features must be accounted for in fully integrated BCI/AAC systems. Translational research should identify and evaluate the specific features (independent variables) that are controlled and manipulated in attempts to improve performance on desired metrics (dependent variables). Figure 1, which has been modified from previous publications,2,7 shows features that BCI researchers should be accountable for identifying when using BCIs as AAC systems.7-9 Primary features, which relate to how language is represented and generated, include language representation methods, access to vocabulary, and methods of utterance generation. Detailed descriptions of the characteristics of primary features have been published elsewhere.10,11 Briefly, the 3 main language representation methods include alphabet-based methods, single meaning pictures, and semantic compaction. Even though multiple language representation methods are available to most AAC speakers, BCI research has focused on the alphabet-based methods of spelling and word prediction.12-14 Two categories of vocabulary words are generally considereddhigh-frequency, core vocabulary words and contextualized, extended vocabulary words. Methods of utterance generation include spontaneous novel utterance generation and pre-stored sentences. Access to core vocabulary maximizes the potential for spontaneous novel utterance generation and allowing AAC speakers to participate in conversation.15 These language-based features should be considered a higher priority than technology features.2,8,10
BCI/AAC systems in the ICF framework The second critical issue to address is integrating BCI/AAC systems into the ICF framework. The ICF framework emphasizes the classification and assessment of functioning and disability in everyday activities, and participation for individuals with medical conditions.16,17 The international AAC community has a diverse group of stakeholders who are multilingual, multicultural,17 and interdisciplinary. The overall aim of the ICF is to provide a unified and standard framework for the description of health and health-related
List of abbreviations: AAC ALS BCI ICF
augmentative and alternative communication amyotrophic lateral sclerosis brain-computer interface International Classification of Functioning, Disability and Health
www.archives-pmr.org
S9 states.5 Using the ICF can help establish uniform terminology for description, assessment, and intervention in the AAC field and improve communication between stakeholders, disciplines, and countries.18-21 The ICF framework is an interactive classification of healthrelated domains, including functioning and disability and contextual factors. Three major components are associated with functioning and disability: (1) body functions and structures, (2) activities, and (3) participation. Two contextual factors influence and are influenced by the functioning and disability components: (1) environmental factors and (2) personal factors. Communication skills can be impacted by any of the major ICF components and can be influenced by a range of environmental and personal factors. Communication has been identified as a common thread across several areas of the ICF including Learning and Applying Knowledge, Interpersonal Interactions and Relationships, and Community, Social, and Civic Life.22 AAC is vital for helping people with complex communication needs communicate. For example, persons with ALS23-25 may have lost the ability to speak but still be involved in daily living activities in the home and community. Their level of participation or ability to hold a conversation, relate stories, or make requests may depend on use of an AAC system, which is classified in the ICF as communication devices and techniques.26 The outcomes achieved using this AAC system are influenced by environmental factors such as communication partner support or a system operator to set up the device. Environmental factors that influence communication outcomes include products and technology for communication, support and relationships, attitudes and services, and systems and policies. Personal factors impacting performance may include motivation and acceptance of the BCI/ AAC system. Consequently, a BCI/AAC system has the potential to integrate with all areas of the ICF framework, as shown in figure 2.
Language-based assessment and intervention Language-based assessment and intervention can be implemented when BCI/AAC systems are considered in the ICF framework. The ICF framework provides a rationale for AAC practice: to improve interactive communication so that an individual’s social participation can be increased.21 This central premise is reflected in the BCI/AAC Language-Based Assessment and Intervention Model.27 This model (fig 3) reflects priorities of foundational research that are required to transform BCIs into fully functional BCI/AAC systems that may be recommended for intervention. To establish strong language-based intervention, foundational factors focused on language and communication should be addressed and supported with evidence before higher-level factors are addressed. The first element of a strong foundation is the goal of BCI/AAC: the most effective interactive communication possible. As reflected in the ICF model, full interpersonal communication is crucial to enhancing one’s potential for education, employment, and independence. Building on this goal, language models and domains are considered when weighing decisions about primary system features. Particular attention may be given to decisions about language representation methods because different language representation methods correspond to fundamentally different approaches to organizing vocabulary content. Objective and informative measures of language performance should be
S10
K. Hill et al
Fig 1 Primary, secondary, and tertiary features of a fully translated BCI/AAC system, including essential BCI device components, as defined by Sellers.9
gathered while a person uses a BCI/AAC system to provide critical evidence for informing the AAC decisionmaking process. Establishing this language-based foundation allows clinicians to consider BCI/AAC technologies in the context of the patient’s communication needs and anticipated levels of performance. This includes matching system features (eg, noninvasive vs invasive BCI; simultaneous vs rapid serial visual presentation of binary stimuli) to an individual’s abilities (cognitive-linguistic skills, vision, hearing, etc) and calibrating the system for optimal performance. Finally, training and ongoing therapy should be implemented with the help of speech-language pathologists and other practitioners. Evidence-based practice during treatment should yield the most effective results and lead to successful, maintained use.
Performance measurement The final critical issue to consider involves directly measuring patients’ communication using a BCI/AAC system to monitor changing skills and ensure that the patient benefits from intervention. The collection and analysis of language samples is the most authentic procedure for identifying communication competence.28,29 Various language sampling tools and resources are available to support systematic observation of an AAC speaker’s
communication competence during and after intervention. Automated performance monitoring provides quantitative data that can be used to evaluate an AAC speaker’s communication competence.30 Built-in language activity monitors used for data logging on several AAC systems, language sample analysis software, and other tools provide effective and efficient methods for collecting language samples and monitoring gains in performance.31 The AAC team is responsible for identifying the most reliable and valid method of collecting spoken and written language samples for analysis. A variety of language sampling contexts may be used for collecting a representative example of an individual’s language functioning and BCI/AAC system use. In the ICF framework,5 the most representative samples would be obtained from language generated during participation in activities of daily living. Capturing these data is not possible without automated data logging. Various language sampling contexts have been used with BCI/AAC speakers, including (1) copy spelling; (2) interview tasks; (3) daily conversation; and (4) e-mail.6 Other language sampling contexts have been reported for different AAC speaker populations, such as picture description tasks.32 Use of multiple sampling contexts allows for comparisons of communication performance across contexts in the summary measures identified above. Metrics used to measure communication competence are similar across the lifespan and well documented. These include
www.archives-pmr.org
Critical issues using brain-computer interfaces
Fig 2
Interactions between the components of ICF and BCI/AAC system.
performance measures used to investigate the language of people who speak verbally and performance measures that are of specific relevance to the population of AAC speakers.33 Measures of typical language performance relate to the subsystems of language (semantics, morphology, syntax) and include standard and defined measures for vocabulary, syntactic diversity, and the length and complexity of utterances. Measures that specifically relate to AAC system use include frequency of spelling, word prediction or other language representation methods, accuracy, average and peak communication rates, selection rates, and rate index. Researchers also investigate trends in the use of the various application programs that are available on a speaker’s AAC system to monitor usage patterns.6
Fig 3
www.archives-pmr.org
S11
Clinical demonstration project The CSP #567 study in the Department of Veterans’ Affairs Healthcare System was a clinical demonstration of an electroencephalography-based BCI for patients with ALS.6 This study is contributing initial evidence on the impact of BCI use for communication in home environments, including evidence on subjects’ ability to generate spontaneous, novel utterances using BCI systems.34 This initial study began to address some of the critical issues raised in the current article. The Wadsworth P300 BCI systema used in this study included alphabet-based methods of language representation (spelling and word prediction) as primary
BCI/AAC Language-Based Assessment and Intervention Model.
S12 features. Tertiary support included training subjects to a criterion level of initial performance using their systems and providing ongoing maintenance, technical support, and further training as needed over an 18-month period. Environmental and personal factors, as defined in the ICF framework, were considered on a case-by-case basis to support subjects’ independent device use in their home environments across diverse levels of participation. A language activity monitor installed on the subjects’ BCI systems was used to collect time-stamped logfiles as a record of language generated using the BCI systems. These logfiles were analyzed to obtain objective measures of communication performance, including word-based and utterance-based performance measures. Interrater reliability for this transcription and analysis process has been reported.35 Fifteen subjects generated at least 1 language sample while using the e-mail or WordPadb program on their BCI system for communication. These subjects were adult veterans with an El Escorial Lab Supported Probable or more definite diagnosis of ALS36 who had lost the ability to communicate either verbally or in writing as indicated by a score of 0 on item 1 or 4 of the ALS Functional Rating Scale, Revised.37 At the time of enrollment, all subjects had a life expectancy of at least 1 year, had corrected visual acuity of at least 20/80, were able to read and understand sixth grade English text on a computer screen, were able to communicate nonverbally, and able to give informed consent using their existing communication methods. During a screening and training phase, all subjects demonstrated sufficient electroencephalographic interaction for the BCI system to operate. This was operationally defined as achieving an average classification accuracy 70% during an elicited copy spelling task and demonstrating independent BCI use 2 times over the course of the screening and training phase. Aggregate results of subjects’ independent BCI use for communication over the course of the study are reported in table 1. These results show that subjects spent a total of 369 hours independently using their BCI systems for communication in their home environments and generated many spontaneous novel utterances in the process. This demonstrates an exciting potential for patients to use BCI systems for expressive communication in realworld settings outside the highly constrained laboratory environments used in experimental studies. However, a limitation of this study was that language-based assessment and intervention services were provided by independent clinicians outside the scope of the research project. Current BCI/AAC systems, including the systems used in this initial study, have been designed for laboratory-based research protocols and are not ready for broad clinical implementation. More work is needed to inform the development and implementation of systems that support the most effective communication possible for BCI speakers. The subjects in this initial study achieved an average communication rate of .43 words per minute. Although communication rates of AAC speakers vary because of factors such as motor abilities (selection rate), use of rate enhancement methods, and interface design, the literature reports an average communication rate of 10 to 15 words per minute for direct keyboard selection.32,38 Systematic investigation of primary, secondary, and tertiary features of BCI/AAC systems may inform the development of improved BCI/AAC systems that allow for more effective interactive communication and faster communication rates.
K. Hill et al Table 1 Summary of BCI language samples in the CSP#567 research study Measure
Unit
Result
Total samples* Total communication timey Total utterances Total words Mean length of utterance Frequency core vocabulary usex Average communication ratejj
N Hours N N Morphemesz % of total words Words per minute
408 369 1212 4764 4.36 54% 0.43
NOTE. Values in table 1 were computed in the Pittsburgh VAMC Laboratory. * Samples include 210 samples of naturalistic e-mail program use, 171 samples of naturalistic WordPad program use, and 27 semistructured interviews completed using WordPad. y Time spent using e-mail and WordPad programs on the BCI device for communication. z Minimal units of meaning. x Percentage of total word tokens that can be found on an established list of high-frequency vocabulary words. jj Average communication rate is a weighted mean found from a subset of 151 samples (661 total utterances) where mean length of utterance >4.0 morphemes.
Future translational research In order for BCIs to be considered as AAC systems, the technology must compete with the existing range of AAC systems, which may be considered as possible ways to achieve effective communication. To accomplish this aim, evidence useful for evaluating a fully functional BCI/AAC system should be available for (1) identifying the features generally used to match a person with BCI/AAC technology; (2) fitting a BCI/AAC system within the ICF framework; (3) implementing language-based clinical services; and (4) measuring and monitoring communication performance. Testing and demonstration projects moving BCI/AAC systems from the bench to the bedside8,39 have not systematically applied and tested the various primary, secondary, and tertiary features that make AAC systems unique, complex rehabilitation and assistive technology. Basic BCI science has focused on secondary features at the level of the control interface to improve signal acquisition and quality in order to improve accuracy. This goal has also led to manipulations of the BCI user interface, such as presentation of arrays with various numbers of locations, letter arrangements, and color options. BCI arrays typically provide access to spelling and word prediction with limited availability of graphic symbols. Providing users with access to multiple language representation methods offers opportunities for comparing the frequency of use of the various language representation methods and the communication rates achieved using specific language representation methods. Tertiary features that are considered when selecting AAC systems include available funding, technical support, maintenance and repair, training, warranties, and practitioners to deliver clinical services. Valued service features have been identified and offered to various degrees of user satisfaction for commercial and fundable AAC systems. BCI/AAC research and development will need to address these features before
www.archives-pmr.org
Critical issues using brain-computer interfaces intervention in a language-based model can be implemented. At least 1 case study has addressed tertiary features by reporting details of training, maintenance and repair, and ongoing technical support that were provided for a long-term BCI user in a home setting.40 Clearly, opportunities for transforming current BCI devices into fully integrated AAC systems are ripe for investigating. Communication is a critical factor for participating in a wide range of activities listed in the ICF. A well-implemented BCI/ AAC intervention may help patients improve their ability to participate in these activities. Translational BCI research may help identify contextual factors that affect the potential for successful BCI/AAC intervention. For example, providing systematic training for system operators and clinical service providers, and providing reliable pathways for technical support may help establish environments that are more likely to encourage successful BCI/AAC system use. Clinicians may also consider personal factors on a case-by-case basis, which may be highly variable across patients. Such factors include the values, beliefs, and expectations of patients and families, acceptance of medical conditions, and other psychosocial factors related to disability. The CSP #567 study6 establishes a baseline for BCI communication performance during written communication in the home and sets a precedent for performance measurement. One would expect translational BCI studies to provide the field with metrics typical of research reporting results of language sample analysis, such as total number of words and utterances, number of different word roots, frequency of core vocabulary, and vocabulary use patterns. The goal of AAC intervention is to achieve the most effective interactive communication possible, as indicated in the first step of the AAC Language-Based Assessment and Intervention Model. Professional organizations such as the American Speech-Language-Hearing Association and the International Society for Augmentative and Alternative Communication expect AAC interventions to result in increased successful participation in activities of daily living for AAC speakers. Intervention using BCI/AAC systems can achieve this goal only if translational research efforts establish a strong languagebased foundation to support clinical practice and provide the means of objectively measuring the language performance of BCI/AAC speakers.
Conclusions BCIs are in transition from laboratory-tested control interfaces to an emerging access option for fully functional BCI/AAC systems. Systematically studying how BCI features and communication contexts affect communication performance and user perception of satisfaction and effectiveness offers limitless opportunities for progress toward the AAC goal of optimized communication for people who cannot speak.41 BCI/AAC studies that use performance data to guide practice will foster BCI systems becoming more mature and functional BCI/ AAC systems.
Suppliers a. Wadsworth P300 BCI system. b. WordPad; Microsoft Corp.
www.archives-pmr.org
S13
Keywords Amyotrophic lateral sclerosis; Communication aids for disabled; Brain-computer interfaces; Language; Rehabilitation; Translational medical research
Corresponding author Katya Hill, PhD, CCC-SLP, Associate Professor, Communication Science and Disorders, University of Pittsburgh, 6017 Forbes Tower, Pittsburgh, PA 15260. E-mail address: [email protected].
Acknowledgments We thank the VA CSP #567 study group identified in appendix 1.
Appendix 1 Participants in the VA CSP #567 Study Participants Study Chair: Robert Ruff, MD, PhD, Louis Stokes Cleveland VAMC; cochairs: Richard Bedlack, Jr, MD, Durham VAMC; Jonathan Wolpaw, MD, Wadsworth Center; chairs office Louis Stokes Cleveland VAMC: Patricia G. Banks, RN, MSNEd, CCRP. Hines VA CSP Coordinating Center: Domenic Reda, PhD, director; Gideon Bahn, PhD, study biostatistician; Kush Kapur, PhD, study biostatistician; Helen Shi, MS, study biostatistician; Tamara Paine, DBA, CCCRP, project manager; Rodney Brown, AA, statistical assistant; Lizy Thottapurathu, statistical programmer; Dyana Gregory, statistical programmer; Mary Biondic, data programmer; Kevin Stroupe, economist; Mary Ellen Vitek, quality assurance specialist; Kwan Hur, study biostatistician: Xue Li, PhD, statistical programmer. CSPCRPCC, Albuquerque, NM: Mike Sather, PhD, center director; Robert Ringer, PharmD, study pharmacist; Dave Hunt, pharmaceutical project manager; Brandi Del Curto, pharmaceutical project manager. Executive Committee: Robert Ruff, MD, PhD, Richard Bedlack, Jr, MD, PhD, Jonathan R. Wolpaw, MD, Katya Hill, PhD, CCC-SLP, Robert Ringer, PharmD, BCNP, Kush Kapur, PhD, Helen Shi, MS, Gideon Bahn, PhD, Patricia Banks, RN, MSNEd, CCRP, Domenic Reda, PhD (ex-officio), Tamara Paine, DBA, CCRP. Data Monitoring Committee: Stephen S. Schreiber, MD, Dennis Mosier, MD, Carolyn Wiles-Higdon, PhD, Samuel S. Wu, PhD. Consultants: Eric W. Sellers, PhD, William Goldberg, PhD, Mary Ellen Vitek, Xue Li, MS, Derrick Kaufman, MS, Linda K. Ganzini, MD, MPH, Pauline Sieverding, MPA, JD, PhD, Thomas Moritz, MS, James Tulsky, MD, Thomas Wingard, Paul Truland. Participating VA Medical Centers: Albany: Donald Higgins, Jr, MD, site investigator; Kimberly Hinman, site coordinator; JoAnn Finn, RN, site coordinator. Cleveland: Robert L. Ruff, MD, PhD, site investigator; Patricia G. Banks, RN, MSNEd, CCRP, national coordinator; Patricia Kouns, site coordinator; Randall Giles, site coordinator; Scott Smith, research assistant. Durham: Richard Bedlack, Jr, MD, site investigator; Beverly McCraw, site coordinator. Providence: Albert Lo, MD, PhD, site investigator; Rosalind Mandelbaum, site coordinator; Susan Fasoli, ScD, OTR/L,
S14 site co-investigator; Milena Gianfrancesco, research assistant; Tara Patterson, PhD, site coordinator. West Haven: Huned Patwa, MD, site investigator; Babar Khokhar, MD, MBA, site coinvestigator; Matthew Laclair, site coordinator. Wadsworth Center Laboratory: Jonathan R. Wolpaw, MD, Susan Heckman, MS, OTyL, Steve Carmack, Eric Sellers, PhD, Dennis J. McFarland, PhD, Sophia Pallone, Stephan Winden, Daniel Corda. Pittsburgh VAMC Laboratory: Katya Hill, PhD, Thomas Kovacs, MA, Sangeun Shin, MS.
References 1. Lloyd LL, Fuller DR, Arvidson HH. Augmentative and alternative communication: a handbook of principles and practices. 1st ed. Boston: Allyn & Bacon; 1997. 2. Hill K. Advances in augmentative and alternative communication as quality-of-life technology. Phys Med Rehabil Clin N Am 2010;21: 43-58. 3. Hill K, Romich B. New clinical tools and methods to support AAC evidence-based practice. In: Proceedings of Center on Disabilities Technology and Persons With Disabilities Conference; 2002 Mar 18e23; Los Angeles, CA. Los Angeles: CSUN; 2002. Available at: http://www.csun.edu/cod/conf/2002/proceedings/191.htm. Accessed July 30, 2013. 4. Hill K, Huggins J. Brain-computer interfaces, ready or not [CDROM]. RESNA 2011 Annual Conference. Toronto: RESNA; 2011. 5. WHO. International Classification of Functioning, Disability and Health: ICF. Geneva: WHO; 2001. 6. Ruff RL, Wolpaw J, Bedlack R. A clinical demonstration of an EEG brain- computer interface for ALS patients. Cooperative study program (CSP) #567, Dept of Veterans Affairs. Clinical trial registration number: NCT00786032; 2011. 7. Hill K, Scherer M. Matching persons and technology: data-driven AAC assessments [CD-ROM]. The 23rd Annual International Conference on Technology and Persons with Disabilities. Los Angeles: CSUN; 2008. 8. Hill K. Augmentative and alternative communication (AAC) research and development: the challenge of evidence-based practice. Int J Comput Process Oriental Lang 2006;19:249-62. 9. Sellers EW. Clinical applications of brain-computer interface technology. Clin EEG Neurosci 2011;42:iv-v. 10. Hill K, Romich B, Vanderheiden G. Augmentative communication. In: Bronzino JD, editor. The biomedical engineering handbook. 4th ed. Boca Raton: CRC Pr; 2010. 11. Cooper R, Ohnabe H, Hobson D. Introduction to rehabilitation engineering. London: Institute of Physics Publishing; 2006. 12. Townsend G, Sellers EW, LaPallo BK, et al. A novel P300-based brain-computer interface stimulus presentation paradigm: moving beyond rows and columns. Clin Neurophysiol 2010;121:1109-20. 13. Mak JN, McFarland DJ, Vaughan TM, et al. EEG correlates of P300based brain-computer interface (BCI) performance in people with amyotrophic lateral sclerosis. J Neural Eng 2012;9:026014. 14. Pires G, Nunes U, Castelo-Branco M. Comparison of a row-column speller vs. a novel lateral single-character speller: assessment of BCI for severe motor disabled patients. Clin Neurophysiol 2012;123: 1168. 15. Hill K, Dollaghan C. Conversations of three-year olds: implications for AAC outcomes. Presented at: American Speech-LanguageHearing (ASHA) Convention; 1999 Nov 18-21; San Francisco, CA. 16. Bilbao A, Kennedy C, Chatterji S, Ustu¨n B, Barquero JLV, Barth JT. The ICF: applications of the WHO model of functioning, disability and health to brain injury rehabilitation. NeuroRehabilitation 2003;18: 239-50.
K. Hill et al 17. Fried-Oken M, Granlund M. AAC and ICF: a good fit to emphasize outcomes. Augment Altern Commun 2012;28:1-2. 18. Huer MB, Hill K, Loncke F. AAC: an international community, an international field, and an international market. Presented at: Convention of RESNA; 2006 Jun 24; Atlanta, GA. Arlington: RESNA; 2006. 19. Huer MB, Hill K, Loncke F. International policy trends and AAC. Biennial Conference of the International Society for Augmentative and Alternative Communication (ISAAC); 2006 Aug 1; Dusseldorf, Germany. Toronto: ISAAC; 2006. 20. Ma EP, Worrall L, Threats TT. The International Classification of Functioning, Disability and Health (ICF) in clinical practice [CDROM]. Semin Speech Lang 2007;28:241-3. 21. Raghavendra P, Bornman J, Granlund M, Bjorck-Akesson E. The World Health Organization’s International Classification of Functioning, Disability and Health: implications for clinical and research practice in the field of augmentative and alternative communication. Augment Altern Commun 2007;23:349-61. 22. Threats TT. Towards an international framework for communication disorders: use of the ICF. J Commun Disord 2006;39: 251-65. 23. Fried-Oken M, Fox L, Rau M, et al. Purposes of AAC device use for persons with ALS as reported by caregivers. Augment Altern Commun 2006;22:209-21. 24. Beukelman DR, Fager S, Nordness A. Communication support for people with ALS. Neurol Res Int 2011;2011:714693. 25. Beukelman DR, Mirenda P. Augmentative and alternative communication: supporting children and adults with complex communication needs. Baltimore: Paul H. Brookes; 2013. 26. Pennington L, Marshall J, Goldbart J. Describing participants in AAC research and their communicative environments: guidelines for research and practice. Disabil Rehabil 2007;29:521-35. 27. Hill K. Augmentative and alternative communication decisions. American Speech-Language-Hearing Association. Available at: http:// www.asha.org/public/speech/disorders/CommunicationDecisions.htm. Accessed July 30, 2013. 28. Light JC, Binger C. Building communicative competence with individuals who use augmentative and alternative communication. Baltimore: Paul H. Brookes; 1998. 29. Hill K, Corsi V. The role of speech language pathologists in assistive technology assessments. In: Scherer MJ, Federici S, editors. Assistive technology assessment: a handbook for professionals in disability, rehabilitation and health professions. London: Taylor & Francis; 2012. p 301-36. 30. Hill K, Romich B. A language activity monitor to support AAC evidence-based clinical practice. Assist Technol 2001;13:12-22. 31. Hill K. Augmentative and alternative communication and language: evidence-based practice and language activity monitoring. Top Lang Disord 2004;24:18-30. 32. Hill K. The development of a model for automated performance measurement and the establishment of performance indices for augmented communicators under two sampling conditions [dissertation]. Pittsburgh: Univ of Pittsburgh; 2001. 33. Hill K. AAC outcomes and performance measurement to support clinical decisions [CD-ROM]. The 15th Biennial of ISAAC; 2012 Jul 30eAug 2; Pittsburgh, PA. 34. Hill K, Kovacs T, Shin S. Are BCIs coming of age as AAC systems? The 5th International Brain-Computer Interface Meeting; 2013 Jun 3e 7; Pacific Grove, CA. International BCI Meeting; 2013. Available at: http://dx.doi.org/10.3217/978-3-85125-260-6-16. 35. Hill K, Kovacs T, Shin S. Reliability of brain computer interface language sample transcription procedures. J Rehabil Res Dev; in press. 36. Brooks BR. El Escorial World Federation of Neurology criteria for the diagnosis of amyotrophic lateral sclerosis. J Neurol Sci 1994; 124(Suppl):96-107.
www.archives-pmr.org
Critical issues using brain-computer interfaces 37. Cedarbaum JM, Stambler N, Malta E, et al. The ALSFRS-R: a revised ALS functional rating scale that incorporates assessments of respiratory function. J Neurol Sci 1999;169: 13-21. 38. Beukelman DR, Mirenda P. Augmentative and alternative communication: supporting children and adults with complex communication needs. Baltimore: P.H. Brookes; 2005.
www.archives-pmr.org
S15 39. Vaughan TM, McFarland DJ, Schalk G, et al. The Wadsworth BCI Research and Development Program: at home with BCI. IEEE Trans Neural Syst Rehabil Eng 2006;14:229-33. 40. Sellers EW, Vaughan TM, Wolpaw JR. A brain-computer interface for longterm independent home use. Amyotroph Lateral Scler 2010;11:449-55. 41. American Speech-Language-Hearing Association. Code of ethics (revised). ASHA Lead 2001;6:2.
SPECIAL COMMUNICATION
Critical Issues Using Brain-Computer Interfaces for Augmentative and Alternative Communication Katya Hill, PhD, CCC-SLP,a,b Thomas Kovacs, MA,a,b Sangeun Shin, MSa,b From aResearch Services, Pittsburgh VA HealthCare System, Pittsburgh, PA; and bAAC Performance, Testing and Teaching Lab, Communication Science and Disorders, School of Health and Rehabilitation Sciences, University of Pittsburgh, Pittsburgh, PA.
Abstract Brain-computer interfaces (BCIs) may potentially be of significant practical value to patients in advanced stages of amyotrophic lateral sclerosis and locked-in syndrome for whom conventional augmentative and alternative communication (AAC) systems, which require some measure of consistent voluntary muscle control, are not satisfactory options. However, BCIs have primarily been used for communication in laboratory research settings. This article discusses 4 critical issues that should be addressed as BCIs are translated out of laboratory settings to become fully functional BCI/AAC systems that may be implemented clinically. These issues include (1) identification of primary, secondary, and tertiary system features; (2) integrating BCI/AAC systems in the World Health Organization’s International Classification of Functioning, Disability and Health framework; (3) implementing language-based assessment and intervention; and (4) performance measurement. A clinical demonstration project is presented as an example of research beginning to address these critical issues. Archives of Physical Medicine and Rehabilitation 2015;96(3 Suppl 1):S8-15 ª 2015 by the American Congress of Rehabilitation Medicine
Augmentative and alternative communication (AAC) is a field of endeavor dedicated to providing effective and efficient interventions, strategies, and technology to help individuals whose natural speech is not functional to participate in their daily activities.1 The most effective AAC interventions supplement or replace speech production, or both, with strategies and technology that offer characteristics of a natural language; for example, the AAC system allows for spontaneous, novel utterance generation.2 Spontaneous novel utterance generation allows a person to formulate sentences and say anything at any time,3 and thus achieve interactive communication. Communication generated using AAC systems can be measured using traditional measures of language performance to gauge overall effectiveness in addition to measures of achieved functional outcomes. Additionally, AAC system features can be selected and
Presented to the National Institutes of Health, National Science Foundation, and other organizations (for a full list, see http://bcimeeting.org/2013/sponsors.html), June 3-7, 2013, Asilomar Conference Grounds, Pacific Grove, CA. Supported in part by the National Institutes of Health (grant no. R123-DE-01274401). The BCI study used as an example in this article was the VA CSP #567 project, funded by the VA Cooperative Studies Program, Department of Veterans Affairs Office of Research and Development. Specific values are reported by the Pittsburgh VAMC Laboratory. Disclosures: none.
manipulated to improve the communication performance of the AAC speaker or system. Brain-computer interfaces (BCIs) may potentially be of significant practical value to patients in advanced stages of amyotrophic lateral sclerosis (ALS) and locked-in syndrome for whom conventional AAC systems, which require some measure of consistent voluntary muscle control, are not satisfactory options. In these individuals, brain signals (eg, P300 potentials) might be good alternatives to channel as a noninvasive approach for accessing assistive technologies. However, measuring the effectiveness of BCIs to support communication becomes an expectation as BCIs mature into fully functional BCI/AAC systems.4 As translational research efforts move BCIs out of laboratory settings to be recommended by speech-language pathologists as viable BCI/AAC systems, quantitative language and communication metrics will become critical for making informed decisions about effectiveness and value. BCI research and development and clinical AAC practice benefit from the use of clearly defined or standardized measures for comparing performance between and within AAC speakers and systems. The purpose of this article is to introduce 4 critical issues that the BCI field should be accountable for addressing as BCI systems mature into BCI/AAC systems that may be implemented
0003-9993/14/$36 - see front matter ª 2015 by the American Congress of Rehabilitation Medicine http://dx.doi.org/10.1016/j.apmr.2014.01.034
Critical issues using brain-computer interfaces clinically. These issues include (1) identification of primary, secondary, and tertiary features of BCI/AAC systems; (2) integrating BCI/AAC systems in the World Health Organization’s International Classification of Functioning, Disability and Health (ICF) framework5; (3) implementing language-based assessment and intervention; and (4) performance measurement. A clinical demonstration project measuring the communication performance of BCI/AAC system speakers6 is presented as an example of research beginning to address these critical issues.
Critical issues to address Primary, secondary, and tertiary device features The first critical issue to address relates to identifying features of BCI/AAC systems. AAC interventions are compared and matched to an individual based on primary, secondary, and tertiary features. These features must be accounted for in fully integrated BCI/AAC systems. Translational research should identify and evaluate the specific features (independent variables) that are controlled and manipulated in attempts to improve performance on desired metrics (dependent variables). Figure 1, which has been modified from previous publications,2,7 shows features that BCI researchers should be accountable for identifying when using BCIs as AAC systems.7-9 Primary features, which relate to how language is represented and generated, include language representation methods, access to vocabulary, and methods of utterance generation. Detailed descriptions of the characteristics of primary features have been published elsewhere.10,11 Briefly, the 3 main language representation methods include alphabet-based methods, single meaning pictures, and semantic compaction. Even though multiple language representation methods are available to most AAC speakers, BCI research has focused on the alphabet-based methods of spelling and word prediction.12-14 Two categories of vocabulary words are generally considereddhigh-frequency, core vocabulary words and contextualized, extended vocabulary words. Methods of utterance generation include spontaneous novel utterance generation and pre-stored sentences. Access to core vocabulary maximizes the potential for spontaneous novel utterance generation and allowing AAC speakers to participate in conversation.15 These language-based features should be considered a higher priority than technology features.2,8,10
BCI/AAC systems in the ICF framework The second critical issue to address is integrating BCI/AAC systems into the ICF framework. The ICF framework emphasizes the classification and assessment of functioning and disability in everyday activities, and participation for individuals with medical conditions.16,17 The international AAC community has a diverse group of stakeholders who are multilingual, multicultural,17 and interdisciplinary. The overall aim of the ICF is to provide a unified and standard framework for the description of health and health-related
List of abbreviations: AAC ALS BCI ICF
augmentative and alternative communication amyotrophic lateral sclerosis brain-computer interface International Classification of Functioning, Disability and Health
www.archives-pmr.org
S9 states.5 Using the ICF can help establish uniform terminology for description, assessment, and intervention in the AAC field and improve communication between stakeholders, disciplines, and countries.18-21 The ICF framework is an interactive classification of healthrelated domains, including functioning and disability and contextual factors. Three major components are associated with functioning and disability: (1) body functions and structures, (2) activities, and (3) participation. Two contextual factors influence and are influenced by the functioning and disability components: (1) environmental factors and (2) personal factors. Communication skills can be impacted by any of the major ICF components and can be influenced by a range of environmental and personal factors. Communication has been identified as a common thread across several areas of the ICF including Learning and Applying Knowledge, Interpersonal Interactions and Relationships, and Community, Social, and Civic Life.22 AAC is vital for helping people with complex communication needs communicate. For example, persons with ALS23-25 may have lost the ability to speak but still be involved in daily living activities in the home and community. Their level of participation or ability to hold a conversation, relate stories, or make requests may depend on use of an AAC system, which is classified in the ICF as communication devices and techniques.26 The outcomes achieved using this AAC system are influenced by environmental factors such as communication partner support or a system operator to set up the device. Environmental factors that influence communication outcomes include products and technology for communication, support and relationships, attitudes and services, and systems and policies. Personal factors impacting performance may include motivation and acceptance of the BCI/ AAC system. Consequently, a BCI/AAC system has the potential to integrate with all areas of the ICF framework, as shown in figure 2.
Language-based assessment and intervention Language-based assessment and intervention can be implemented when BCI/AAC systems are considered in the ICF framework. The ICF framework provides a rationale for AAC practice: to improve interactive communication so that an individual’s social participation can be increased.21 This central premise is reflected in the BCI/AAC Language-Based Assessment and Intervention Model.27 This model (fig 3) reflects priorities of foundational research that are required to transform BCIs into fully functional BCI/AAC systems that may be recommended for intervention. To establish strong language-based intervention, foundational factors focused on language and communication should be addressed and supported with evidence before higher-level factors are addressed. The first element of a strong foundation is the goal of BCI/AAC: the most effective interactive communication possible. As reflected in the ICF model, full interpersonal communication is crucial to enhancing one’s potential for education, employment, and independence. Building on this goal, language models and domains are considered when weighing decisions about primary system features. Particular attention may be given to decisions about language representation methods because different language representation methods correspond to fundamentally different approaches to organizing vocabulary content. Objective and informative measures of language performance should be
S10
K. Hill et al
Fig 1 Primary, secondary, and tertiary features of a fully translated BCI/AAC system, including essential BCI device components, as defined by Sellers.9
gathered while a person uses a BCI/AAC system to provide critical evidence for informing the AAC decisionmaking process. Establishing this language-based foundation allows clinicians to consider BCI/AAC technologies in the context of the patient’s communication needs and anticipated levels of performance. This includes matching system features (eg, noninvasive vs invasive BCI; simultaneous vs rapid serial visual presentation of binary stimuli) to an individual’s abilities (cognitive-linguistic skills, vision, hearing, etc) and calibrating the system for optimal performance. Finally, training and ongoing therapy should be implemented with the help of speech-language pathologists and other practitioners. Evidence-based practice during treatment should yield the most effective results and lead to successful, maintained use.
Performance measurement The final critical issue to consider involves directly measuring patients’ communication using a BCI/AAC system to monitor changing skills and ensure that the patient benefits from intervention. The collection and analysis of language samples is the most authentic procedure for identifying communication competence.28,29 Various language sampling tools and resources are available to support systematic observation of an AAC speaker’s
communication competence during and after intervention. Automated performance monitoring provides quantitative data that can be used to evaluate an AAC speaker’s communication competence.30 Built-in language activity monitors used for data logging on several AAC systems, language sample analysis software, and other tools provide effective and efficient methods for collecting language samples and monitoring gains in performance.31 The AAC team is responsible for identifying the most reliable and valid method of collecting spoken and written language samples for analysis. A variety of language sampling contexts may be used for collecting a representative example of an individual’s language functioning and BCI/AAC system use. In the ICF framework,5 the most representative samples would be obtained from language generated during participation in activities of daily living. Capturing these data is not possible without automated data logging. Various language sampling contexts have been used with BCI/AAC speakers, including (1) copy spelling; (2) interview tasks; (3) daily conversation; and (4) e-mail.6 Other language sampling contexts have been reported for different AAC speaker populations, such as picture description tasks.32 Use of multiple sampling contexts allows for comparisons of communication performance across contexts in the summary measures identified above. Metrics used to measure communication competence are similar across the lifespan and well documented. These include
www.archives-pmr.org
Critical issues using brain-computer interfaces
Fig 2
Interactions between the components of ICF and BCI/AAC system.
performance measures used to investigate the language of people who speak verbally and performance measures that are of specific relevance to the population of AAC speakers.33 Measures of typical language performance relate to the subsystems of language (semantics, morphology, syntax) and include standard and defined measures for vocabulary, syntactic diversity, and the length and complexity of utterances. Measures that specifically relate to AAC system use include frequency of spelling, word prediction or other language representation methods, accuracy, average and peak communication rates, selection rates, and rate index. Researchers also investigate trends in the use of the various application programs that are available on a speaker’s AAC system to monitor usage patterns.6
Fig 3
www.archives-pmr.org
S11
Clinical demonstration project The CSP #567 study in the Department of Veterans’ Affairs Healthcare System was a clinical demonstration of an electroencephalography-based BCI for patients with ALS.6 This study is contributing initial evidence on the impact of BCI use for communication in home environments, including evidence on subjects’ ability to generate spontaneous, novel utterances using BCI systems.34 This initial study began to address some of the critical issues raised in the current article. The Wadsworth P300 BCI systema used in this study included alphabet-based methods of language representation (spelling and word prediction) as primary
BCI/AAC Language-Based Assessment and Intervention Model.
S12 features. Tertiary support included training subjects to a criterion level of initial performance using their systems and providing ongoing maintenance, technical support, and further training as needed over an 18-month period. Environmental and personal factors, as defined in the ICF framework, were considered on a case-by-case basis to support subjects’ independent device use in their home environments across diverse levels of participation. A language activity monitor installed on the subjects’ BCI systems was used to collect time-stamped logfiles as a record of language generated using the BCI systems. These logfiles were analyzed to obtain objective measures of communication performance, including word-based and utterance-based performance measures. Interrater reliability for this transcription and analysis process has been reported.35 Fifteen subjects generated at least 1 language sample while using the e-mail or WordPadb program on their BCI system for communication. These subjects were adult veterans with an El Escorial Lab Supported Probable or more definite diagnosis of ALS36 who had lost the ability to communicate either verbally or in writing as indicated by a score of 0 on item 1 or 4 of the ALS Functional Rating Scale, Revised.37 At the time of enrollment, all subjects had a life expectancy of at least 1 year, had corrected visual acuity of at least 20/80, were able to read and understand sixth grade English text on a computer screen, were able to communicate nonverbally, and able to give informed consent using their existing communication methods. During a screening and training phase, all subjects demonstrated sufficient electroencephalographic interaction for the BCI system to operate. This was operationally defined as achieving an average classification accuracy 70% during an elicited copy spelling task and demonstrating independent BCI use 2 times over the course of the screening and training phase. Aggregate results of subjects’ independent BCI use for communication over the course of the study are reported in table 1. These results show that subjects spent a total of 369 hours independently using their BCI systems for communication in their home environments and generated many spontaneous novel utterances in the process. This demonstrates an exciting potential for patients to use BCI systems for expressive communication in realworld settings outside the highly constrained laboratory environments used in experimental studies. However, a limitation of this study was that language-based assessment and intervention services were provided by independent clinicians outside the scope of the research project. Current BCI/AAC systems, including the systems used in this initial study, have been designed for laboratory-based research protocols and are not ready for broad clinical implementation. More work is needed to inform the development and implementation of systems that support the most effective communication possible for BCI speakers. The subjects in this initial study achieved an average communication rate of .43 words per minute. Although communication rates of AAC speakers vary because of factors such as motor abilities (selection rate), use of rate enhancement methods, and interface design, the literature reports an average communication rate of 10 to 15 words per minute for direct keyboard selection.32,38 Systematic investigation of primary, secondary, and tertiary features of BCI/AAC systems may inform the development of improved BCI/AAC systems that allow for more effective interactive communication and faster communication rates.
K. Hill et al Table 1 Summary of BCI language samples in the CSP#567 research study Measure
Unit
Result
Total samples* Total communication timey Total utterances Total words Mean length of utterance Frequency core vocabulary usex Average communication ratejj
N Hours N N Morphemesz % of total words Words per minute
408 369 1212 4764 4.36 54% 0.43
NOTE. Values in table 1 were computed in the Pittsburgh VAMC Laboratory. * Samples include 210 samples of naturalistic e-mail program use, 171 samples of naturalistic WordPad program use, and 27 semistructured interviews completed using WordPad. y Time spent using e-mail and WordPad programs on the BCI device for communication. z Minimal units of meaning. x Percentage of total word tokens that can be found on an established list of high-frequency vocabulary words. jj Average communication rate is a weighted mean found from a subset of 151 samples (661 total utterances) where mean length of utterance >4.0 morphemes.
Future translational research In order for BCIs to be considered as AAC systems, the technology must compete with the existing range of AAC systems, which may be considered as possible ways to achieve effective communication. To accomplish this aim, evidence useful for evaluating a fully functional BCI/AAC system should be available for (1) identifying the features generally used to match a person with BCI/AAC technology; (2) fitting a BCI/AAC system within the ICF framework; (3) implementing language-based clinical services; and (4) measuring and monitoring communication performance. Testing and demonstration projects moving BCI/AAC systems from the bench to the bedside8,39 have not systematically applied and tested the various primary, secondary, and tertiary features that make AAC systems unique, complex rehabilitation and assistive technology. Basic BCI science has focused on secondary features at the level of the control interface to improve signal acquisition and quality in order to improve accuracy. This goal has also led to manipulations of the BCI user interface, such as presentation of arrays with various numbers of locations, letter arrangements, and color options. BCI arrays typically provide access to spelling and word prediction with limited availability of graphic symbols. Providing users with access to multiple language representation methods offers opportunities for comparing the frequency of use of the various language representation methods and the communication rates achieved using specific language representation methods. Tertiary features that are considered when selecting AAC systems include available funding, technical support, maintenance and repair, training, warranties, and practitioners to deliver clinical services. Valued service features have been identified and offered to various degrees of user satisfaction for commercial and fundable AAC systems. BCI/AAC research and development will need to address these features before
www.archives-pmr.org
Critical issues using brain-computer interfaces intervention in a language-based model can be implemented. At least 1 case study has addressed tertiary features by reporting details of training, maintenance and repair, and ongoing technical support that were provided for a long-term BCI user in a home setting.40 Clearly, opportunities for transforming current BCI devices into fully integrated AAC systems are ripe for investigating. Communication is a critical factor for participating in a wide range of activities listed in the ICF. A well-implemented BCI/ AAC intervention may help patients improve their ability to participate in these activities. Translational BCI research may help identify contextual factors that affect the potential for successful BCI/AAC intervention. For example, providing systematic training for system operators and clinical service providers, and providing reliable pathways for technical support may help establish environments that are more likely to encourage successful BCI/AAC system use. Clinicians may also consider personal factors on a case-by-case basis, which may be highly variable across patients. Such factors include the values, beliefs, and expectations of patients and families, acceptance of medical conditions, and other psychosocial factors related to disability. The CSP #567 study6 establishes a baseline for BCI communication performance during written communication in the home and sets a precedent for performance measurement. One would expect translational BCI studies to provide the field with metrics typical of research reporting results of language sample analysis, such as total number of words and utterances, number of different word roots, frequency of core vocabulary, and vocabulary use patterns. The goal of AAC intervention is to achieve the most effective interactive communication possible, as indicated in the first step of the AAC Language-Based Assessment and Intervention Model. Professional organizations such as the American Speech-Language-Hearing Association and the International Society for Augmentative and Alternative Communication expect AAC interventions to result in increased successful participation in activities of daily living for AAC speakers. Intervention using BCI/AAC systems can achieve this goal only if translational research efforts establish a strong languagebased foundation to support clinical practice and provide the means of objectively measuring the language performance of BCI/AAC speakers.
Conclusions BCIs are in transition from laboratory-tested control interfaces to an emerging access option for fully functional BCI/AAC systems. Systematically studying how BCI features and communication contexts affect communication performance and user perception of satisfaction and effectiveness offers limitless opportunities for progress toward the AAC goal of optimized communication for people who cannot speak.41 BCI/AAC studies that use performance data to guide practice will foster BCI systems becoming more mature and functional BCI/ AAC systems.
Suppliers a. Wadsworth P300 BCI system. b. WordPad; Microsoft Corp.
www.archives-pmr.org
S13
Keywords Amyotrophic lateral sclerosis; Communication aids for disabled; Brain-computer interfaces; Language; Rehabilitation; Translational medical research
Corresponding author Katya Hill, PhD, CCC-SLP, Associate Professor, Communication Science and Disorders, University of Pittsburgh, 6017 Forbes Tower, Pittsburgh, PA 15260. E-mail address: [email protected].
Acknowledgments We thank the VA CSP #567 study group identified in appendix 1.
Appendix 1 Participants in the VA CSP #567 Study Participants Study Chair: Robert Ruff, MD, PhD, Louis Stokes Cleveland VAMC; cochairs: Richard Bedlack, Jr, MD, Durham VAMC; Jonathan Wolpaw, MD, Wadsworth Center; chairs office Louis Stokes Cleveland VAMC: Patricia G. Banks, RN, MSNEd, CCRP. Hines VA CSP Coordinating Center: Domenic Reda, PhD, director; Gideon Bahn, PhD, study biostatistician; Kush Kapur, PhD, study biostatistician; Helen Shi, MS, study biostatistician; Tamara Paine, DBA, CCCRP, project manager; Rodney Brown, AA, statistical assistant; Lizy Thottapurathu, statistical programmer; Dyana Gregory, statistical programmer; Mary Biondic, data programmer; Kevin Stroupe, economist; Mary Ellen Vitek, quality assurance specialist; Kwan Hur, study biostatistician: Xue Li, PhD, statistical programmer. CSPCRPCC, Albuquerque, NM: Mike Sather, PhD, center director; Robert Ringer, PharmD, study pharmacist; Dave Hunt, pharmaceutical project manager; Brandi Del Curto, pharmaceutical project manager. Executive Committee: Robert Ruff, MD, PhD, Richard Bedlack, Jr, MD, PhD, Jonathan R. Wolpaw, MD, Katya Hill, PhD, CCC-SLP, Robert Ringer, PharmD, BCNP, Kush Kapur, PhD, Helen Shi, MS, Gideon Bahn, PhD, Patricia Banks, RN, MSNEd, CCRP, Domenic Reda, PhD (ex-officio), Tamara Paine, DBA, CCRP. Data Monitoring Committee: Stephen S. Schreiber, MD, Dennis Mosier, MD, Carolyn Wiles-Higdon, PhD, Samuel S. Wu, PhD. Consultants: Eric W. Sellers, PhD, William Goldberg, PhD, Mary Ellen Vitek, Xue Li, MS, Derrick Kaufman, MS, Linda K. Ganzini, MD, MPH, Pauline Sieverding, MPA, JD, PhD, Thomas Moritz, MS, James Tulsky, MD, Thomas Wingard, Paul Truland. Participating VA Medical Centers: Albany: Donald Higgins, Jr, MD, site investigator; Kimberly Hinman, site coordinator; JoAnn Finn, RN, site coordinator. Cleveland: Robert L. Ruff, MD, PhD, site investigator; Patricia G. Banks, RN, MSNEd, CCRP, national coordinator; Patricia Kouns, site coordinator; Randall Giles, site coordinator; Scott Smith, research assistant. Durham: Richard Bedlack, Jr, MD, site investigator; Beverly McCraw, site coordinator. Providence: Albert Lo, MD, PhD, site investigator; Rosalind Mandelbaum, site coordinator; Susan Fasoli, ScD, OTR/L,
S14 site co-investigator; Milena Gianfrancesco, research assistant; Tara Patterson, PhD, site coordinator. West Haven: Huned Patwa, MD, site investigator; Babar Khokhar, MD, MBA, site coinvestigator; Matthew Laclair, site coordinator. Wadsworth Center Laboratory: Jonathan R. Wolpaw, MD, Susan Heckman, MS, OTyL, Steve Carmack, Eric Sellers, PhD, Dennis J. McFarland, PhD, Sophia Pallone, Stephan Winden, Daniel Corda. Pittsburgh VAMC Laboratory: Katya Hill, PhD, Thomas Kovacs, MA, Sangeun Shin, MS.
References 1. Lloyd LL, Fuller DR, Arvidson HH. Augmentative and alternative communication: a handbook of principles and practices. 1st ed. Boston: Allyn & Bacon; 1997. 2. Hill K. Advances in augmentative and alternative communication as quality-of-life technology. Phys Med Rehabil Clin N Am 2010;21: 43-58. 3. Hill K, Romich B. New clinical tools and methods to support AAC evidence-based practice. In: Proceedings of Center on Disabilities Technology and Persons With Disabilities Conference; 2002 Mar 18e23; Los Angeles, CA. Los Angeles: CSUN; 2002. Available at: http://www.csun.edu/cod/conf/2002/proceedings/191.htm. Accessed July 30, 2013. 4. Hill K, Huggins J. Brain-computer interfaces, ready or not [CDROM]. RESNA 2011 Annual Conference. Toronto: RESNA; 2011. 5. WHO. International Classification of Functioning, Disability and Health: ICF. Geneva: WHO; 2001. 6. Ruff RL, Wolpaw J, Bedlack R. A clinical demonstration of an EEG brain- computer interface for ALS patients. Cooperative study program (CSP) #567, Dept of Veterans Affairs. Clinical trial registration number: NCT00786032; 2011. 7. Hill K, Scherer M. Matching persons and technology: data-driven AAC assessments [CD-ROM]. The 23rd Annual International Conference on Technology and Persons with Disabilities. Los Angeles: CSUN; 2008. 8. Hill K. Augmentative and alternative communication (AAC) research and development: the challenge of evidence-based practice. Int J Comput Process Oriental Lang 2006;19:249-62. 9. Sellers EW. Clinical applications of brain-computer interface technology. Clin EEG Neurosci 2011;42:iv-v. 10. Hill K, Romich B, Vanderheiden G. Augmentative communication. In: Bronzino JD, editor. The biomedical engineering handbook. 4th ed. Boca Raton: CRC Pr; 2010. 11. Cooper R, Ohnabe H, Hobson D. Introduction to rehabilitation engineering. London: Institute of Physics Publishing; 2006. 12. Townsend G, Sellers EW, LaPallo BK, et al. A novel P300-based brain-computer interface stimulus presentation paradigm: moving beyond rows and columns. Clin Neurophysiol 2010;121:1109-20. 13. Mak JN, McFarland DJ, Vaughan TM, et al. EEG correlates of P300based brain-computer interface (BCI) performance in people with amyotrophic lateral sclerosis. J Neural Eng 2012;9:026014. 14. Pires G, Nunes U, Castelo-Branco M. Comparison of a row-column speller vs. a novel lateral single-character speller: assessment of BCI for severe motor disabled patients. Clin Neurophysiol 2012;123: 1168. 15. Hill K, Dollaghan C. Conversations of three-year olds: implications for AAC outcomes. Presented at: American Speech-LanguageHearing (ASHA) Convention; 1999 Nov 18-21; San Francisco, CA. 16. Bilbao A, Kennedy C, Chatterji S, Ustu¨n B, Barquero JLV, Barth JT. The ICF: applications of the WHO model of functioning, disability and health to brain injury rehabilitation. NeuroRehabilitation 2003;18: 239-50.
K. Hill et al 17. Fried-Oken M, Granlund M. AAC and ICF: a good fit to emphasize outcomes. Augment Altern Commun 2012;28:1-2. 18. Huer MB, Hill K, Loncke F. AAC: an international community, an international field, and an international market. Presented at: Convention of RESNA; 2006 Jun 24; Atlanta, GA. Arlington: RESNA; 2006. 19. Huer MB, Hill K, Loncke F. International policy trends and AAC. Biennial Conference of the International Society for Augmentative and Alternative Communication (ISAAC); 2006 Aug 1; Dusseldorf, Germany. Toronto: ISAAC; 2006. 20. Ma EP, Worrall L, Threats TT. The International Classification of Functioning, Disability and Health (ICF) in clinical practice [CDROM]. Semin Speech Lang 2007;28:241-3. 21. Raghavendra P, Bornman J, Granlund M, Bjorck-Akesson E. The World Health Organization’s International Classification of Functioning, Disability and Health: implications for clinical and research practice in the field of augmentative and alternative communication. Augment Altern Commun 2007;23:349-61. 22. Threats TT. Towards an international framework for communication disorders: use of the ICF. J Commun Disord 2006;39: 251-65. 23. Fried-Oken M, Fox L, Rau M, et al. Purposes of AAC device use for persons with ALS as reported by caregivers. Augment Altern Commun 2006;22:209-21. 24. Beukelman DR, Fager S, Nordness A. Communication support for people with ALS. Neurol Res Int 2011;2011:714693. 25. Beukelman DR, Mirenda P. Augmentative and alternative communication: supporting children and adults with complex communication needs. Baltimore: Paul H. Brookes; 2013. 26. Pennington L, Marshall J, Goldbart J. Describing participants in AAC research and their communicative environments: guidelines for research and practice. Disabil Rehabil 2007;29:521-35. 27. Hill K. Augmentative and alternative communication decisions. American Speech-Language-Hearing Association. Available at: http:// www.asha.org/public/speech/disorders/CommunicationDecisions.htm. Accessed July 30, 2013. 28. Light JC, Binger C. Building communicative competence with individuals who use augmentative and alternative communication. Baltimore: Paul H. Brookes; 1998. 29. Hill K, Corsi V. The role of speech language pathologists in assistive technology assessments. In: Scherer MJ, Federici S, editors. Assistive technology assessment: a handbook for professionals in disability, rehabilitation and health professions. London: Taylor & Francis; 2012. p 301-36. 30. Hill K, Romich B. A language activity monitor to support AAC evidence-based clinical practice. Assist Technol 2001;13:12-22. 31. Hill K. Augmentative and alternative communication and language: evidence-based practice and language activity monitoring. Top Lang Disord 2004;24:18-30. 32. Hill K. The development of a model for automated performance measurement and the establishment of performance indices for augmented communicators under two sampling conditions [dissertation]. Pittsburgh: Univ of Pittsburgh; 2001. 33. Hill K. AAC outcomes and performance measurement to support clinical decisions [CD-ROM]. The 15th Biennial of ISAAC; 2012 Jul 30eAug 2; Pittsburgh, PA. 34. Hill K, Kovacs T, Shin S. Are BCIs coming of age as AAC systems? The 5th International Brain-Computer Interface Meeting; 2013 Jun 3e 7; Pacific Grove, CA. International BCI Meeting; 2013. Available at: http://dx.doi.org/10.3217/978-3-85125-260-6-16. 35. Hill K, Kovacs T, Shin S. Reliability of brain computer interface language sample transcription procedures. J Rehabil Res Dev; in press. 36. Brooks BR. El Escorial World Federation of Neurology criteria for the diagnosis of amyotrophic lateral sclerosis. J Neurol Sci 1994; 124(Suppl):96-107.
www.archives-pmr.org
Critical issues using brain-computer interfaces 37. Cedarbaum JM, Stambler N, Malta E, et al. The ALSFRS-R: a revised ALS functional rating scale that incorporates assessments of respiratory function. J Neurol Sci 1999;169: 13-21. 38. Beukelman DR, Mirenda P. Augmentative and alternative communication: supporting children and adults with complex communication needs. Baltimore: P.H. Brookes; 2005.
www.archives-pmr.org
S15 39. Vaughan TM, McFarland DJ, Schalk G, et al. The Wadsworth BCI Research and Development Program: at home with BCI. IEEE Trans Neural Syst Rehabil Eng 2006;14:229-33. 40. Sellers EW, Vaughan TM, Wolpaw JR. A brain-computer interface for longterm independent home use. Amyotroph Lateral Scler 2010;11:449-55. 41. American Speech-Language-Hearing Association. Code of ethics (revised). ASHA Lead 2001;6:2.
