Do You Hear What I Hear? Reflecting on Auditory Display in Medicine

• Abstract
• Introduction
• Why use auditory display?
• Language of auditory display
• Designing auditory displays
• Ineffective use of auditory display in the medical environment.
• Effective auditory display in health informatics
• Telehealth
• Anaesthesiology
• Medical Records and Computerised Prescription Order Entry Systems
• Patient Selection
• Patient-Prescription Mismatch
• Conclusion
• Acknowledgments
• References

Abstract
The use of auditory display in medical environments could reduce errors associated with interfacing with a computerised health system. Auditory displays must be designed effectively, accounting for aspects of cognitive work to be useful and effective in complementing current technology. Audible alarms designed to alert medical personnel to specific problems have been ineffective as the association between the alarm and the error have often been difficult to identify. We seek to explore the possibilities for effective auditory display that could allow for quick identification of medical issues and reduce the errors associated with computer interfaces in the medical environment. The auditory display paradigm must evaluate problems faced in the past and develop new methods of displaying data with sound. We explore areas of auditory display in telehealth, anaesthesiology, and medical records as a means to increasing effectiveness and reducing mental workload.

Introduction
Unless the interface of a computerised health system is designed to integrate all the necessary information to ensure effective interpretation and usage, full acceptance of computerised systems in medicine will not be achieved. Human–machine interface problems can result in incorrect medication dosage, patient-prescription mismatch, and inadequate data management.^[1–5] Alarms alerting the user to potential problems are often disregarded, especially in situations of prior "false alarms".^[1] Classical interface designs have often failed to fully satisfy the needs of the health care community.

Medical environments are gradually becoming dependent on information available from machines. Computer systems are responsible for monitoring vital signs and administering drugs in addition to collecting data and maintaining records. As a result, the information displayed by these machines, via the user interface, must be understandable in high stress environments. Many cluttered displays are present in a given environment, especially the operating theatre or the intensive care unit. Medical personnel must be able to continuously scan information presented on a variety of devices critical to the care of the patient. Information may be missed if users are only relying on visual means rather than capitalising on other senses.

Auditory information can enable rapid identification of patient care issues or machine malfunction. Auditory displays (as opposed to auditory alarms) can offer an advantage over visual displays. An auditory component can either be used to complement the visual display or replace the visual display. An auditory display is one that transforms data into sound. Displays where sound would be advantageous are those in which there is a high cognitive load with critical requirement for operator vigilance, data sets that are hard to display, or data that cannot be displayed visually.^[6] We seek to explore the characteristics of auditory displays that can be used in a medical environment to reduce the types of errors that have been observed to date. Methods for designing and improving auditory displays will also be discussed.

Why use auditory display?
As with visual display, auditory display is not without its advantages and disadvantages. The methods to determine the benefits of auditory displays and how to apply auditory interfacing will be addressed later in this article.

With sound, transient events that may otherwise be missed are easily identified. For example, visual electroencephalogram (EEG) displays must be monitored concurrently with other systems. The potential to miss a sudden change in a patient’s condition is much greater in a visual environment than in an auditory environment where an increase or decrease in pitch or tempo is quickly noticed. Multiple sounds can be used to display information about many variables. Using sound enables a better understanding of changing information and less confusion than would be required using multiple graphical variables.^[7]

The use of sound can elicit an automatic response and is omnidirectional, not requiring specific orientation to relay information. A doctor can attend to a patient rather than continuously monitor a screen. Also, sound display reduces the physical space required for visual displays in front of the doctor: a series of computer monitors does not need to be mounted in a given configuration to enable quick response. Although sound can be heard without localising, there are situations in which sound can be used to draw attention to a specific area or device, thus providing information about location. However, Weinger cites difficulties prioritising alarms as determining the sound source can be challenging.^[8] By increasing the variability or spectral content of the sound and using representative sounds for each machine (a heart beat for pulse monitoring), the auditory display can support quick localisation of the source.

In a computing environment with a singular display, auditory localisation is achieved with directional cues. These cues can be provided using interaural time differences (ITD – a different part of the wave was presented to each ear dependent on the length of the wave) and interaural intensity differences (IID – intensity at each ear depended on the position from the obstacle). This draws the visual attention to the same area as that of the auditory tone. It can draw the attention to a specific visual focal point.

In cases of high mental workload or significant data display, the information relayed through audition could reduce stress on medical personnel. Brewster has performed several studies that have shown that auditory interfacing can decrease mental workload while performing specific computer tasks.^{[9–12]} Tasks such as clicking on buttons (or icons), following menu tasks, and copying and moving files can be performed faster and with less mental workload when auditory information is used to compliment visual information.

Another use of sound for medical environments is to provide alternate channels for information when several visual interfaces are used simultaneously or when the visual interface is small. Personal digital assistants (PDAs) are slowly becoming popular among medical personnel for reporting purposes and for accessing vital information at the bedside.^[13] Auditory information can reduce the need for the screen itself to display information. In some environments, PDAs can be used to enable diabetics to monitor their own insulin levels.^[14] One of the side effects of diabetes is often loss of vision. Individuals with low vision can capitalise on auditory information.

Although in many cases, the advantages outweigh the disadvantages, those environments that employ teams of individuals that may be exposed to the auditory display must be considered. The first issue is that of annoyance. As sound is omnidirectional and cannot be avoided by looking to another part of the display, annoyance is a real possibility. Annoyance has been a factor in the lack of effective use of the current alarm systems in medical environments. These alarms have not sought to improve the display as a whole, but rather to indicate problems. Sound can be used as a channel for environmental information providing cues about the welfare of the patient. With experience in listening to auditory displays, the sounds become background noise which only alerts the caregiver if the sound itself changes. For example, if the sound of a heartbeat is used to convey pulse and the heartbeats increase in tempo, the caregiver realises that the pulse is increasing. Mapping the sounds to specific data allows the user to easily identify problems. By using sound to enhance information within the environment, annoyance may be reduced. Brewster has found in several studies that an informational auditory display does not increase annoyance as compared with visual displays.^[10-12]

Not only can sound provide information to the user of the system but it can also provide information in a collaborative environment, eg, to allow operating room staff to work together in understanding the needs of the patient. There are two types of sound that can be used to convey information: speech and non-speech sounds. Speech refers to combinations of sound that work together to convey a specific message. Non-speech sounds are tones that are mapped to data and must be learned. Speech information is more difficult to process than non-speech signals. Usually, medical personnel communicate using speech. A surgeon could respond automatically to an auditory display signalling patient distress, rather than waiting for verbal communication from the anaesthesiologist. Critical timing scenarios can be worked through without the added stress of communication. Not only is this important for operating room personnel who are used to working together, but those who may work in teams, both locally or internationally, with whom they are not familiar.

Prior to implementing sound displays in an environment, careful consideration must be given to the design of the auditory display to ensure complementary technology. Next we will discuss the language of auditory display, followed by a discussion of the methods by which auditory interface design can be achieved.

Language of auditory display
As the elements of context in auditory interface design differ from that of visual interface design, this section discusses the terms used in the auditory interface domain. As with visual displays that use icons, there are forms of reference for auditory interfacing. These include auditory icons, earcons, audification and sonification.

Auditory icons represent a sound "image" of the object, or motion to which it is referring. This is a direct comparison to visual icons. A heartbeat sound can be used for monitoring pulse information. The heartbeat is well recognised, especially by physicians, and provides more interpretable information than that of a number representing pulse. Testing has shown that learning and response times are faster for auditory icons than for other signals such as speech warning.^[15] This supports the collaborative work environment in that all medical personnel in a given environment can respond quickly.

Earcons are abstract musical tones that are usually represented in hierarchical form to relay information. These tones must be learned by the user in order to be useful in interpreting the information they represent. Earcons can be used in combinations to display multiple meanings, in the same manner that phrases are a combination of words. Earcons are currently used for display of obstacle information for individuals with visual impairments. For instance, the Sonic Pathfinder is used as an obstacle alerting device.^[16] As an individual approaches an obstacle, the pitch of the device increases along a chromatic scale.

Audification represents direct translation of physical energy into audible sound. This is similar to echolocation that visually impaired individuals use to orient themselves. An individual provides a "click" or raps a cane and the echoes reflect back to the individual to indicate the distance to obstacles in the room.

Lastly, sonification is the mapping of data streams onto auditory dimensions. The use of sonification to detect sleep apnoea has been proposed by Ballora.^[17] This system maps heart rate variability to sound. Each interbeat interval is mapped to a specific pitch. As the heart rate increases, the pitch also increases. To notify the medical personnel of larger interbeat intervals (larger than 50 msec), a timbral annotation or "tinkling" sound is heard. This auditory mapping complements the visual display and can be used to better understand how the frequency of heart rate oscillations sounds in a given environment.

Individual elements of context can be used to represent a given environment, but additional information can also be added. Much the same as multivariate icons can be created in visual displays, sounds can also be multivariate in their display. Methods to combine information in a given sound include changing different parameters of the sound. These include changing the tempo, pitch, timbre, amplitude and spatial cues of the tone. Tempo can be used in an iconic form to give a sense of urgency. Increased pitch gives the perception of being louder, thus perceptually, pitch and loudness are often related. Timbre is the sound quality of complex sounds. Finally spatial cues can provide information about where to turn, or where to look to attend to audible information. By resampling, changing the pitch or compressing the energy of EEG data, information can be translated to the audible range. Jovanov has proposed the use of sonification in EEG activity to improve perception of global patterns of brain activity over time.^[18] The system that they propose enables a 3D spatialisation based on EEG information. The pitch in this case corresponds to the drowsiness of the patient and alerts the neurophysiologist or EEG technician.

These available elements of context offer rich displays that provide a significant amount of information especially in combination with visual displays. Careful consideration must be given to the design of the auditory display to ensure that the information is presented in an effective manner and to reduce the possibility of information overload.

Designing auditory displays
Considerable thought must be given to the design of the auditory display before determining which sounds to use. The users of the device, and the environment in which the system will be used, must be given due consideration. Also, the ability to revisit information that has been previously presented and the amount of information that can be reviewed is critical (looking back through visual records or graphs is fairly easy, whereas this is not necessarily a simple task for information transmitted with sound).^[19] By taking into account various aspects of the cognitive work analysis, the display can be designed to construe important information to the user without causing information overload.

The first stage of the cognitive work analysis is that of ecological interface design (EID). This has been shown to be useful in medical environments, aviation and process control. EID is a systematic approach to the design of interfaces for complex systems.^[14] Using work domain analysis in the form of a five-level hierarchy, from the concrete description of the system components to an abstract evaluation of the overall system purpose, the critical relationships can be recognised. Additionally, cognitive work analysis allows for reducing the mental workload of integrating, calculating, or remembering multiple values.^[13] Several areas of medical care have made use of EID, including nursing, anaesthesia monitoring and diabetic patient monitoring.^[13] EID for diabetic patient monitoring has involved the development of simplified interfaces usable on PDAs. Blood glucose levels are shown relative to the ideal (too high, too low, normal) and historical data for previous days is also readily available. Self-monitoring support systems for diabetic patients is one area where EID can be used highly effectively.

EID is typically applied to a visual environment, but can be used in combination with other cognitive analyses such as strategy analysis, social organisation analysis, attentional mapping and worker competencies analysis in the design of auditory environments.^[6] Strategy analysis aids in identifying the different methods of doing tasks identified in the earlier analyses. Redundancy in auditory systems can be accomplished by matching visual and auditory techniques to allow for prompt detection. Social organisation analysis involves the determination of work sharing. Buxton used earcons in identifying user tasks in collaborative environments.^[20] This assessment can enable feedback not only about tasks within the domain of the individuals themselves but about the tasks in the environmental domain. A surgeon can be aware of the patient’s anaesthesiology needs while maintaining his/her concentration on the task at hand. The specific tasks requiring the auditory and visual focus is important to determine in the system and can be established with the aid of attentional mapping. Finally, the worker competencies analysis enables detection of those tasks which are skill (instinctive), rule (routine) or knowledge-based (reasoning or problem solving). These behaviours determine the information that needs to be supplied to enable a response by the user.

By combining the use of different methods of cognitive work analysis, a good understanding of the requirements of the system and the collaborative work environment can be achieved. Once these have been identified, the methods to achieve these requirements through various elements of context can be employed. It is important to note that, although all these cognitive requirements are defined, design is an iterative process. The design of the prototype interface must be tested, revised and retested until the users of the system are satisfied that additional improvements are no longer required.

This method of test–retest does not appear to have been conducted in the past for auditory displays. As a result, alarm systems have become unpopular in hospital environments. The lack of user-aided design has resulted in the failure of these alarms to alert individuals who often ignore or turn off the alarms.^[8] The next section will discuss the ineffective use of alarms in the current medical environment.

Ineffective use of auditory display in the medical environment.
Unfortunately, the majority of auditory displays in the present medical environment only provide alarms. Alarms are forms of symbolic auditory display, but do not provide additional information to the user about the nature of the situation to which they refer. Alarms have been designed for the purpose of fault management, but the misunderstanding of alarm signals has prevented their effective use. In some cases, alarms have been misjudged resulting in life-threatening scenarios. Weinger cites 11 human factor issues for auditory alarms in health care.^[8] These can be summarised in two broad categories. The first category covers inadequate training of personnel and their inability to interpret alarms. Within this category, different sounds may require different responses depending on the environment. The same device may be used at home or in the ICU (intensive care unit) setting. Alarms that are effective at home may be insignificant in the ICU environment. The second category covers the design of the alarm itself. There are similar alarms for different situations and different types of equipment. These design flaws result in an inability to prioritise the alarms. There are also false alarms which wrongly alarm the medical personnel. By providing more useful displays, rather than just alarms, identification of the problem can more easily be ascertained.

Research has shown that symbolic alarms are ineffective at alerting medical staff to problems in anaesthesiology. Finley and Cohen found that anaesthetists had difficulty recognising alarms in an operating room and were not likely to respond.^[21] Another study performed by the National Patient Safety Foundation,^[22] found that when auditory alarms are used, the time required for laryngoscopy and intubation increased. ISO standards were developed to enhance the response to alarms such that those with a higher priority would be more easily interpreted. Mondor and Finley performed a study of 13 alarms on machines that met the ISO standards used in the operating room including the Datex AS/3 monitor and the Hewlett Packard model 66s monitor to determine the urgency of the sounds by novice participants.^[23] They found that the perceived urgency was not reflected by the sound mapped to the event. Those that were indicative of patient condition appeared less urgent than those dealing with equipment malfunction. Even the ISO standards do not offer sufficient guidance on effective use of auditory display. The design must employ auditory display information in a meaningful manner in order for it to become accepted in the health domain.

Effective auditory display in health informatics
There are several environments in which auditory display is used and proposed for use to complement those systems that have in the past been primarily visual. These areas take into account the importance of using auditory display to relay information in a manner that may be more understandable by the medical communities. These include videoconferencing, anaesthesiology and, finally, general practice.

Telehealth
Telehealth is an area that depends on information from the auditory environment especially when evaluating patients in environments that are not easily physically accessible by care providers (eg, the northern parts of Canada). The Northern Telehealth Network in Saskatchewan, eg, uses audible information to perform routine check ups for obstetrical patients and evaluating patients recovering from heart surgery from a distance. At the basic level, sound is used to provide a communication portal between physician and patient using interactive video and audio. In this case, doctors can perform consults and examinations without being physically located at the northern clinic. At a higher level, information is also being transferred through telehealth systems. More specifically, electrocardiography uses auditory display as well as visual. Canada has many telehealth networks to communicate with individuals in far northern environments. In Quebec, a study found that using telehealth to evaluate electrocardiograms (ECGs) remotely eliminated 70 percent of the trips (elective visits or transfers) which would otherwise have been taken to a major centre.^[24]

Most telehealth systems convey ECGs directly through the ECG plot, but this information can be improved through sonification. Ballora modelled ECG information with sonification to identify four heart conditions: good health, congestive heart failure, atrial fibrillation and sleep apnoea.^[25] Sonifications included higher heart rates represented by higher pitch and interbeat intervals greater than 50 msec as a change in timbre. Using this model of sonification, 30 novice users were able to differentiate between healthy heart conditions and sleep apnoea. This method of sonification may provide auditory information that is otherwise difficult to evaluate.

Anaesthesiology
Another area that auditory signals have been suggested as useful for in the medical realm is anaesthesiology. Watson et al^[26] indicate that tailoring of alarms in an operating room is a complex task which many anaesthetists fail to do. As mentioned earlier, many alarms are turned off in the operating room or ignored. Their recommendation is that auditory displays be designed that assist the anaesthetist to better identify patent status and needs. Watson and Sanderson have studied the use of sonification to provide additional information to anaesthetists.^[27] One study of respiratory monitoring used musical tones to represent inhalation and exhalation and tempo to display the respiratory rate. As the sound intensity ramped up, the tidal volume could be determined and, as pitch changed, the amount of carbon dioxide could be determined. This type of sonification was found to be effective by anaesthetists in clinical environments at maintaining their level of performance while improving their ability to perform alternate tasks.

Medical Records and Computerised Prescription Order Entry Systems
Auditory display is now used in many computer programs. Windows®, for instance, allows for sounds to be turned on for critical battery situations, USB connections or disconnections, new mail, menu pop ups and many other exclamations of interest. Each of these is a different sound, iconic in nature. Most represent the sound of the information they are trying to portray. When a program is being opened, the sound of a door opening is heard and when closing a file, a door is slammed shut. For many computer environments, the use of sound is researched in significant detail to enable users a method to recognise errors and provide useful information without alarming the individual. This is an area of research virtually untouched in the world of electronic health records (EHRs) and physician order entry systems.

Sound could also be used to reduce the mental workload while a physician is multitasking in a medical environment. Vision provides information about a small area in a select environment whereas sound can provide several levels of information without the requirement to concentrate on any one level. For example, in a typical environment, a physician will be conversing with a patient while recording information about the client’s current needs. They will also be reviewing the patient’s file for evidence of allergies, reactions and current medications. While printing a prescription for the individual, the physician may also be determining the correct dose to prescribe, mentally evaluating any other drug interactions while ensuring that the client is aware of risks and side effects. The mental workload to perform these operations may be reduced by introducing auditory displays.

Compound hierarchical earcons can be developed to allow significant information to be relayed without the need for concentration on one specific aspect.^[28] Brewster has shown that software tasks can be accomplished with a lesser mental workload using auditory information from specialised earcons.^[9-11] These include sonic enhancement of visual icons (buttons), scrollbars, menus and drag and drop facilities.

Patient Selection
Errors associated with patient selection in EHR systems or computerised prescription order entry (CPOE) systems are believed to be a result of small font size. One area in which auditory display might reduce this effect is the ability to use dictation to indicate patient names. The doctor or nurse would be able to say the name and the selection would appear. This could immediately bring up the file of the individual patient. Dictation is also used to transcribe information and is the only area currently considered by vendors of EHR systems. Dragon Dictate software is included in some systems to allow doctors to dictate the patient information into the file. Nurses have been found to be accepting of this form of data recording.^[29]

Patient-Prescription Mismatch
Errors that have been noted in CPOE systems and EHRs include difficulties in patient selection, mis-selection of medications, incorrect medication dosage, patient prescription mismatch and inadequate data management. Many of these errors are a result of incorrect selection of menu items. The result is a slip off the menu item which results in either an incorrect selection or the need to repeat the task. This issue is notable not only in health care environments and has been studied in computing environments by Brewster and Crease.^[10]

Following the Brewster and Crease recommendations for increasing menu usability, we can reduce workload in menu selection tasks, with the intention of reducing error.^[10] Assume a pull-down menu that requires selection of patient, then selection of medication and, finally, selection of dosage. A variety of sounds can be used to represent each set of requirements. Earcons representing related menu items can be displayed to indicate that the mouse or cursor is located within a menu. Then, if the sound changes, a menu slip has occurred. For instance, a family of organ tones could be used for each task.^[10] Organ tones have a wide range of available pitches allowing for increased variability in menu options. Earcons for patient selection could start with a percussive organ. As each patient name is highlighted, alternate pitches could be heard. If the physician slid off the menu, the feedback sound would stop. If he slipped to a different patient, a change in pitch would alert him to the change. If a selection were made, a two-tiered tone of 40 msec at a higher intensity would be heard. The same approach could be used for medication selection, this time with a pipe organ, and finally the dosage could be a synthetic organ.^[10]

Conclusions
Auditory displays are commonly used by to enhance the use of software without detracting from the visual display. This trend is only now being extended to the medical environment. By employing auditory displays in a computing environment, one can reduce the cognitive workload of the user. It is believed that by incorporating effective auditory information into current systems for electronic patient monitoring and physician order entry, the reduction of errors may occur.

Acknowledgements
We wish to acknowledge the insightful suggestions made by Dominic Covvey. This work was supported by the Canadian Institutes of Health Research.

References

1. Abookire SA, Teich JM, Sandige H, Paterno MD, Martin MT, Kuperman GJ et al. Improving allergy alerting in a computerized physician order entry system. Proceedings / AMIA 2000;Annual:Symposium-6.
2. Ash JS, Berg M, Coiera E. Some unintended consequences of information technology in health care: the nature of patient care information system-related errors. J Am Med Inform Assoc 2004 Mar 1;11(2):104–12.
3. Payne TH, Nichol WP, Hoey P, Savarino J. Characteristics and override rates of order checks in a practitioner order entry system. Proceedings / AMIA 2002;Annual:Symposium-6.
4. Koppel R, Metlay JP, Cohen A, Abaluck B, Localio AR, Kimmel SE et al. Role of computerized physician order entry systems in facilitating medication errors. [See comment.] JAMA 2005 Mar 9;293(10):1197–203.
5. Shah NR, Seger AC, Seger DL, Fiskio JM, Kuperman GJ, Blumenfeld B et al. Improving acceptance of computerized prescribing alerts in ambulatory care. J Am Med Inform Assoc 2006 Jan 1;13(1):5–11.
6. Sanderson P, Anderson J, Watson M. Extending ecological interface design to auditory displays. ARC Key Centre for Human Factors. Appl Cognitive, Queensland Univ, Australia; 2000. p 259–6.
7. Weinstein JR, Cook PR. FAUST: A framework for algorithm understanding and sonification testing. International Conference on Auditory Display, (ICAD) 1997.
8. Weinger MB. Auditory warnings in the medical work domain: an overview of critical issues. Proceedings of the XIVth Triennial Congress of the International Ergonomics Association and 44th Annual Meeting of the Human Factors and Ergonomics Association, ’Ergonomics for the New Millennium’; Anesthesia Ergonom. Res. Laboratory, Department of Anesthesiology, University of California, San Diego, CA, United States. San Diego, CA, United States: Human Factors and Ergonomics Society, Santa Monica, CA 90406, United States; 2000. p211–4.
9. Brewster SA, Wright PC, Dix AJ, Edwards ADN. The sonic enhancement of graphical buttons. Human-Computer Interaction. Interact ’95; VTT Inf. Technol., Finland. Lillehammer, Norway: Chapman & Hall; 1995. p. 43–8.
10. Brewster SA, Crease MG. Correcting menu usability problems with sound. Behaviour and Information Technology 1999 May;18(3):165–77.
11. Brewster SA. Using non-speech sound to overcome information overload. Displays 1997 May 1;17(3-4):179–89.
12. Brewster SA, Wright PC, Edwards ADN. The design and evaluation of an auditory-enhanced scrollbar. CHI ’94 Conference Proceedings. Human Factors in Computing Systems `Celebrating Interdependence’; Dept. of Comput. Sci., York Univ., UK. Boston, MA, USA: ACM; 1994 p 173–9.
13. Momtahan K, Burns CM. Applications of ecological interface design in supporting the nursing process. J Healthc Inf Manag 2004;18(4):74–82.
14. Burns CM, Hajdukiewicz JR. Ecological interface design. Boca Raton, FL: CRC Press; 2004.
15. Graham R. Use of auditory icons as emergency warnings: evaluation within a vehicle collision avoidance application. Ergonomics 1999;42(9):1233–48.
16. Heyes T. Sonic pathfinder. Electronics & Wireless World 1984 Apr;90(1579):26–9.
17. Ballora M, Pennycook B, Ch Ivanov P, Goldberger A, Glass L. Detection of obstructive sleep apnea through auditory display of heart rate variability. Computers in Cardiology; Pennsylvania State Univ, University Park, PA, USA. Cambridge, MA, USA: Institute of Electrical and Electronics Engineers Computer Society, Los Alamitos, CA, USA; 2000. p 739–40.
18. Jovanov E, Wegner K, Radivojevic V, Starcevic D, Quinn MS, Karron DB. Tactical audio and acoustic rendering in biomedical applications. IEEE Transactions on Information Technology in Biomedicine 1999;3(2):109–18.
19. Sanderson P, Watson. From information content to auditory display with ecological interface design: prospects and challenges. ARC Key Centre for Human Factors & Appl. Cognitive, Queensland Univ, Australia; 2005. p 259–63.
20. Buxton W, Moran TP. EuroPARC’s integrated interactive intermedia facility (IIIF): early experiences. Proceedings of the IFIP WG 8.4 Conference on Multi-user Interfaces and Applications, Heraklion, Crete. Amsterdam: Elsevier Science Publishers B.V. (North-Holland); 1990. pp 11-34.
21. Finley GA, Cohen AJ. Perceived urgency and the anaesthetist: responses to common operating room monitor alarms. Can J Anaesth 1991;38(8):958–64.
22. Xiao Y, Seagull FJ, Mackenzie C, Wickens C, Via DK. Focus patient safety 2002;5(1): 3–5. Ref Type: Magazine Article.
23. Mondor TA, Finley GA. The perceived urgency of auditory warning alarms used in the hospital operating room is inappropriate. [See comment.] Can J Anaesth 2003 Mar;50(3):221–8.
24. Bellavance M, Beland MJ, van Doesburg NH, Paquet M, Ducharme FM, Cloutier A. Implanting telehealth network for paediatric cardiology: learning from the Quebec experience. Cardiology in the Young 2004 Dec;14(6):608–14.
25. Ballora M, Pennycook B, Ch Ivanov P, Glass L, Goldberger A. Heart rate sonification: a new approach to medical diagnosis. Leonardo 2004;37(1):41–6.
26. Watson M, Sanderson P, Russell WJ. Tailoring reveals information requirements: The case of anaesthesia alarms. Interact Comput 2004;16(2):271–93.
27. Watson M, Sanderson P. Sonification supports eyes-free respiratory monitoring and task time-sharing. Hum Factors 2004;46(3):497–517.
28. Blattner MM, Sumikawa DA, Greenberg RM. Earcons and icons: their structure and common design principles. Human-Computer Interaction 1989;4(1):11–44.
29. Dillon TW, Norcio AF. User performance and acceptance of a speech-input interface in a health assessment task. Int J Hum Comput Stud 1997;47(4):591–602.