developing icu-talk. a computer based communication aid for patients in intensive care - 2002

8/12/2019 Developing ICU-Talk. a Computer Based Communication Aid for Patients in Intensive Care - 2002

1/7

Developing ICU-Talk - A Com puter based Comm unication Aidfor Patients in Intensive C are

A Judson', F MacAulay2, M Etchels3, I W Ricketts', A Waller', J Brodie2,A W arden3, N Al m', A Shearer3,B Gordon4.Department of Applied Co mputing, University of Dundee.Speech and Language Therapy Department, Ninewells H ospital, Dundee.Intensive Care U nit, Ninewells Hospital, Dundee.School of Nursin g Midwifery, University of Dundee.

bstractThis pap er highlights the developm ent of a

compu terised communication aid fo r intubated patients inan Intensive Care Unit ICU).Intubated patients in ICU have few methods ofcommunicating other than attempting t o mouth words a nduse gestures. Communication aids that are available, suchas symbol cha rts and alphabet boards, unfortunately havenot been designed to meet the spec@ requirements ofthese patients. E ven these methods are not ideal fo r thepatient , their family members or the sta flcar ing for thepatient. From prev ious research stud ies of communicationwithin ICU, patients often fee l disempow ered anddepressed due their inability to contribute toconversations. This often leads to feel ings of soci lisolation. ICU-Talk will be developed to allow the patientto have more control of their communications and at thesame time address their ongoing problems and needs onan individual basis.

IntroductionThe ICU-Talk project is a three-year collaborativeproject which aims to develop and test a computer basedcommunication aid designed to meet the specific needs ofintubated patients in Intensive Care Units. The project hasbeen funded by the Engineering and Physical ScienceResearch Council (EPSRC). The development team for theICU-Talk system consists of a senior staff nurse from ICU,a speech and language therapist, a software engineer and adoctoral student in computing. The multi-disciplinarynature of this team means that a range of skills andknowledge including clinical experience of patient care,knowledge of human communication patterns, assistive

communication devices, and software/interface designhave been brought together to ensure all these aspects areconsidered in the design, implementation and evaluation ofICU-Talk.

BackgroundThe use of communication aids with patients who areintubated and/or ventilated as in ICU has always beenrestricted. These patients are unable to talk because of atube in their nose, mouth or throat (endotracheal tube)which passes into their windpipe (trachea) and helps withbreathing but temporarily stops the vocal cords fromworking. They are also often very unwell, with acuteillnesses or severe injury following an accident. Many

patients admitted t an ICU only stay for a short time andmay be sedated throughout. However, there is a group ofpatients whose stay in ICU is prolonged. These patientsmay be awake and orientated but unable to communicateeffectively. Weakness in their hands and arms means theyare unable to write or use gestures. This group of patientspresent a particular set of problems as they attempt tocommunicate with health care personnel, friends andrelatives. Their attempts at mouthing words or using facialexpression are often misinterpreted or are simply notunderstood. This situation can lead to feelings offrustration, isolation, stress and inadequacy and can affectthe patient's recovery [2] [3].Currently available Augmentative and AlternativeCommunication (AAC) devices have been designed eitherfor clients with language impairment and/or learningdifficulties or for clients with intact language but aphysical disability. No commercially available AACdevices have been d esigned with the specific needs of theICU patient in mind. The patient's medical condition andthe high levels of stress and anxiety experienced bypatients in ICUs mean they tire very quickly. These factors

1089 6503/0010.00 2000 IEEE 37


2/7

also affect their cognitive skills including their ability tolearn, retain information and concentrate on tasks. This inturn affects the patients ability to use an AAC device.Many patients may also have physical deficits. These caninclude weakness, loss of sensation and reduced movementof the hands, arms, legs, feet and head. This makesaccessing any AAC device problematic, as the patient maynot be able to use a keyboard, mouse, trackball or touchscreen. Finding a reliable movement for use with a singleswitch may be possible but currently available switchoperated AAC systems are either highly cognitively loadedor designed only to convey very basic needs. Given thesechallenges it is not surprising that the use of high-techcomputer based communication aids in ICU has beengenerally unsuccessful. Low-tech communication aids e.g.alphabet charts, picture boards and eye pointing systemsare used along with mouthing, writing, gesture and facialexpression, but these systems are time consuming,frustrating and inaccurate [l].Nursing staff in ICU are highly skilled in facilitatingcommunication with intubated patients. They usecombinations of yeslno questions to try and determinepatients needs and wants and with experience theybecome proficient at anticipating and interpreting what thepatient is attempting to communicate. This worksreasonably well with needs based communication butbecomes much more difficult when the nurse and patientdo not share common knowledge about a subject or if thetopic of conversation is unrelated to the patientsimmediate environment. For patients who haveprolonged stay in ICU this reduction in communicationwill affect their ability to build and maintain relationshipsand social closeness with friends, family, nurses, and othermembers o f the hospital team.

The TalksBac systemAn augmentative communication system, designedspecifically for and tested by adults with dysphasia, wasdeveloped during four years of collaborative researchinvolving Dund ee Speech Language Therapy Serviceand Dundee Universitys Applied Computing Department

[5] This system is called TalksBac. TalksBac allows theuser to access a database of pre-stored sentences andstories. Its interface is designed to be sim ple to operate andnavigation through the system can be achieved by mouse,trackball or touch screen. TalksBac also has the potentialto support a simple scanning interface (see Figure 1).TalksBac uses prediction to help guide the user throughtheir interaction. The transparency of the TalksBac systemmakes it useful for patients who have a short concentrationspan and whose ability to learn and remember iscompromised. It was felt that the design principles used forTalksBac could also be applied to help patients in ICU

communicate more effectively. To asses this a pilotTalksBac database was developed which illustrated theways in which such a system could support a patient inICU. Feedback from the ICU consultant and nursing staffprovided positive support for developing the ICU-Talksystem.

I Figure I Screenshot of TalksBac interface.Creating an initial database of patientutterances

Nursing questionnaireThe ICU nursing staff were interviewed to obtain aninitial list of phrases normally used by ICU patients. Usingexisting clinical expertise the development team was ableto identify nine commonly used topics of conversation:

EnvironmentNeedsFeelingsFamily and friendsWorkMedicalFuture-short termFuture-long termOther

A short structured interview was assembled and thirty-two of the forty-two ICU nursing staff were interviewedover a period of three weeks. Each nurse was asked toprovide examples of three or more, questions or commentsthat patients usually say within a given topic.

371


3/7

For example:Environment Topic- It is too hot or Wheream I Needs Topic- Ineed a drink or I need a coughMedical Topic When will I get y tube out?

The development team combined similar phrases,questions and comments. The responses from the nursingstaff formed a core database of one hundred and ninetyphrases arranged under the nine topics listed above. Sometopics contain more phrases than others. A data recordercontaining these results was developed for the purpose ofobserving patients attempting to commu nicate in ICU.

Patient selectionHaving created the core database, the next phase was toobserve suitable intubated patients in ICU who were tryingto communicate and hereby establish what patients wereactually trying to say (see figure 2).

Figure 2. Patient in ICU attempting tocommunicate with daughter.An ideal patient profile, in the form of a flow chart,was compiled to help nursing staff identify suitablepatients. The flow chart prompts the nursing staff with aseries of questions. If at any point the answer to a questionis no, then the patient is not suitable for inclusion in theproject. The questions asked were:Is the patient intubated?Is the sedation score between 1 and 3?Is the patient able to consistently respond to a basiccommand?Is the patient attempting to commu nicate?Is the patient able to focus?Was he patient literate prior to admission?

Nursing staff will be able to use the flow chart, after theproject has ended, to identify patients who may be able touse a communication aid.The sedation score is a scoring system that is unique toNinewells ICU. An assessment of the patients level ofconsciousness is made and a score of between 1 and 6given to that patient [4]. Patients who are able tocommunicate have a sedation score of between 1 and 3

Sedation scoring system:1. Alert and anxious.2. Aware and calm.3 Drowsy and co-operative.4.5.6. Unconscious and unrousable.

Asleep; responds to verbal command o r lightglabellar tap.Asleep; responds t o painful stimuli.

After referral by the nursing staff, the projectdevelopment team assessed the patient and decided if thepatient was suitable for inclusion within the project. Thepatients permission to participate with the project was alsoobtained. On acceptance, the project aims were explainedto the patient and they received a demon stration of the datarecorder system. Support from the patients family orfriends was sought and they were involved withdiscussions about the project and the expected outcomes.Only phrases communicated by the patient were recordedand the nursing staff, family and friends were made awareof this.

Patient o b s e r v a t i o nA computerised core data recorder was developed to

assist with observation of communication by selectedpatients in ICU. The recorder holds the core database ofphrases provided by the nursing staff and the observermatches the patients communication attempts to these oradds any novel ph rases.The observer records who the patient is trying tocommunicate with e.g. family, nurse, doctor, and howthe patient communicates e.g. mouthing, gesturing, oralphabet board. Then for each patient communication, theobserver selects the topic and the phrase communicated.Each phrase used by a patient is recorded in the observerspersonal event trace file. The trace includes who thepatient spoke to, how they did it, what they said and whenit was said. If the phrase that the patient communicates isnot contained within the database then the observer willadd it under the appropriate topic.

37


4/7

Early analysis indicates that patients say very individualthings but there are common phrases, such as I need acough", I need a drink" and "Thank you".Using this process a increasing representative coredatabase will be compiled containing the most commonlyused patient phrases.

PhrasesA total of 406 phrases were observed and recorded. 178of these were phrases were present in the core database.This represents 25 of the core database. The remaining228 phrases were patient specific. Figure 3shows the 21

most frequently communicated phrases.Results

Wh oThe results detailed below are taken from 12 ICUpatients. These patients were observed for a total of 30hours. The results were compiled from the event trace fileand the database files that the observation software usesand maintains.The person with whom the patient was communicatingwas recorded for each phrase. The most commoncommunication partner was the nurse. The patientcommunicated with the nurse 44 of the time. The nextmost common communication partner was a familymember. Although there are other staff in ICU they

Ineed the suction tubing inmy handIwant to write it down

I'm not nghtWhat are you doing to me

Hold my hand pleaseI hattered

GoodbyeChange positioduncomfortable

YesI eel better

Iwant to go homeIwant to sleep

When w ll Iget my tube outI ore

Iwant to sit upI oo hotI'm okay

HelloThank you

need a drink -wa sh my mouthneed a cough _ _ _ _ ~ _ ~

0 2 4 6 8 10 12 14 16 18 20Frequency of Use

Figure 3. Phrases most frequently comm unicated by patients.

373


5/7

normally do not talk directly to the patient but insteadcommunicate via the patients nurse or relative. Theircommunications are such that they rarely give the patientthe opportunity to ask questions or engage in conversation.

TopicsThe phrases were organised under nine topics. Feelingsand Needs were the topics most frequently used. From theobservation results, the topics future-long term andfuture-short term were combined to form the new topicfuture. The number of possible conversational topics asindicated in the figures below, now contains eight topics.

Figure 4. Standard (boxes) interface.How

The method used by the patient to communicate aphrase is referred to as how. How the patientcommunicated each phrase was also observed andrecorded. If more than one method was used tocommunicate then the predominant method was recorded.Mouthing was used most often by patients (63 ) followedby gesture (25 ). The term gesture was used to describe aphysical gesture or facial expression used in acommunicative way. These results reflect the patientsnatural reaction to try to use speech to communicate eventhough they know they are unable to achieve a voice.Many patients had severe weakness of their upper limbsand were unable to write, produce specific gestures orpoint accurately to an alphabet chart.

Interface designThe design of a suitable interface is central to the ICU-Talk system. Navigation by a patient to select an utteranceneeds to be fast, easy to learn and memorable. Advice oninterface design was sought from a local softwaredevelopment company and principles from computer gamedesign incorporated into ICU-Talk to make it both visuallystimulating and intuitive in use. One of the importantunderlying principles was consistency, e.g. If a section ofthe program was to present displays as sliding from left toright then this had to be continued throughout the program.Similarly orders of buttons and arrows should alsoillustrate the movement of left to right The specific needsof the target patient group also need to be considered ateach stage of the interface development. It was decided

that the patient should be able to choose which interfacethey prefer from a choice of two styles, standard (boxes) ordynamic (bubbles), (see figures 4 and 5). Both interfaceslink to the same underlying algorithms and data structures.

I Figure 5 . Dynamic (bubbles) interface.Patients who will benefit from ICU-Talk are likely to havea combination of physical and cognitive problems [ l ] andthis has influenced the interface design.The ICU-Talk system has been designed so thatminimal training is required and phrases are ordered toallow access with the minimum number of selections. Aspart of the training the patient will be asked to set some ofthe interface parameters. These parameters will include thesensitivity of the touch screen, speed of the mouseemulator and the background colour, font size and style forthe phrase buttons. This means that the patient cancustomise the interface to a limited degree and so havesome control over the appearanc e of the system.When starting each session with ICU-Talk the patient isasked to select how they are feeling from a choice of threeicons that represent their m ood. They then select an energylevel from a choice of six. The energy level represents howtired or how full of energy the patient feels (see figure 6 ) .It is used together with the patients mood selection toallow the system to order phrases e.g. if the patients mood

74


6/7

Private phrases hiddenFigure 6 Mood and energy screen selector.

is happy and combined with high energy levels thenpositive phrases will be placed at the top of the phrase lists.The next selection the patient makes is the topic ofconversation. The interface containing the topic choiceshas easy to read buttons that provide a large target area forselection by touch screen or mouse emulator. The systemalso has the facility for any single-switch user to use ascanning system to navigate the interface.The topic buttons are colour coded so the patient canbuild up an association between the colour and the topic toaid their navigation through the system. Once the topic ischosen a list of potential phrases is displayed (see figure7). The background colour of the phrases screen is the

Figure 7. Phrase selecto r screen.same as the topic colour. The colour of the phrase buttons,the size and style of font o f the phrases is determined bythe patient.The ICU-Talk system also includes the facility to haveprivate phrases hidden from general view. The patientcan select the icon ( a picture of a padlock) on the interface,which will then allow these additional phrases to bedisplayed on the screen. In many commercially availableAAC devices all phrases are displayed on the screen andare potentially available for everyone to see. By keepingprivate phrases hidden from view the patient can have

more control over conversations, and build and maintainrelationships in a more normal way e.g. the patient mayhave things they want to say to their spouse but not to thestaff in ICU.

Patient evaluation of the systemThe iterative nature of the design of the ICU-Talksystem has allowed potential users to view the interfacedesign and to assess input devices and provide feedback onthe how these could be altered or improved. Some of thepatients who have had their communication observed inICU and have viewed the interface designs for ICU-Talk,commented on how easy it is to learn to use and navigate.They have also commented favourably on colour schemes,font type and size. These users have also tested a range ofinput devices and enabled us to confirm that even withlimited control of their upper limbs, they were able to usethem effectively. Input devices assessed included mouse,drag-pad, trackball, as well as a range of single switches.Each patient had their preferred input device. The ability toaccommodate individual patients preferences is anessential component in the development of ICU-Talk. Easyaccessibility to adjust information and alter theperformance of the system is needed to meet the ever-changing nee ds of the patient.

Conclusions and future workDevising an AAC system specifically for patients inICU is particularly challenging due to the patientsindividual and complex needs and the limitations imposedby their condition and the ICU environment. Developing a

database containing phrases that will be of most use tointubated patients who are temporarily unable to speak iscentral to the success of ICU-Talk.Nursing staff from ICU were interviewed to identifythose phrases used most frequently by patients tocommunicate. Patients who were attempting tocommunicate were then observed. Patients used a quarterof the phrases suggested by nursing staff. The majority o fphrases communicated by patients were patient specific,reinforcing the need for an additional database ofpersonalised phrases. This will be achieved with the co-operation of the patients family and friends. A structuredcomputerised interview, which will be used to assemblethe relevant data and integrate it into the database.The majority of phrases communicated by patients wereunder the topics of needs and feelings suggesting thatpatients are most concerned about their presentcircumstan ces and not the future.

375


7/7

The most common communication partner for thepatient is the nurse. Within the ICU, nurses work 12 hourshifts and each nurse responsible for one patient. Thisallows a relationship to be built up between the patient andthe nurse. Other members of staff tend to communicate viathe nurse rather than directly with the patient. It istherefore essential to obtain a perception of the ICU -Talksystem from the nursing staff. This will be achievedthrough q uestionnaires and interviews.Friends and relatives have shared knowledge of thepatient and how the patient tries to communicate tomaintain their relationship and offer reassurance andsupport to the patient while in this unfamiliar andsometimes frightening environment. By communicatingwith friends and relatives the patient is able to keep intouch w ith normal life outside the ICU setting.Patients use mouthing most frequently to communicateeven though they are unable to vocalise. For many patientsthis is the most natural form of communication but it canbe very difficult to lip read patients especially if they areorally intubated. The patient is unable to produce anysound therefore it can be di fic ul t to attract attention, andwhen lip reading the communication partner has to belooking directly at the patient throughout the conversation.ICU-Talk will speak out the pre-stored phrases once theyare selected making conversations m ore natural.The interface design is ongoing and includes evaluationby potential users, nursing staff and relatives/partners ofthe patients. A full prototype system including the abilityto include a patient specific database will be evaluated inICU over a twelve month period and the results from thissecond phase of the project will be reported at futuremeetings.

References[l ] Albarran J W. (1991). A review of communication withintubated patients and those with tracheostomies within anintensive care setting.Intensive Care Nursing. Vol. 7.[2] Costello J . (1998). AAC with the temporarily nonspeakingpatient in the intensive care unit: The Childrens HospitalBoston Model. In: Proceedings of the BiennialConference of the International Societyfor Augmentative andAlternative Communication, Dublin, Ireland, 24-27 August1998 1 897606 04 4), pp 6-7.[3] Leathart A J. (1994). Communication and socialisation( ):anexploratory study and explanation for nurse patientcommunication in an ITU. Intensive and Critical CareNursing. Vol 10[4] Ramsay M A E., Savage T M., Simpson B R J., Goodwin R.(1974) Controlled sedation with Alphaxalone and Alphadone.British Medical Journal:2: pp 656 659

[ ] Waller A., Dennis F., Brodie K., Cairns A Y. (1997).Evaluating the use of TalksBac, a predictive communicationdevice for non-fluent aphasic adults. International Journal ofLanguage and Comm unication Disorders. 33:1, pp 45-70.

cknowledgementsThe team wishes to acknowledge Russell Kay of Visual SciencesLtd, Irene McPhail, John Low and Sandra Hill for their support.The grant holders also thank the UK Engineering and PhysicalSciences Research Council for fund ing the project.

376

developing icu-talk. a computer based communication aid for patients in intensive care - 2002

Documents