A PROPOSAL OF A WORKFLOW-BASED CBR FOR

INTERVENTIONAL CDSS

Application to Transcatheter Aortic Valve Implantation (TAVI)

Andres El-Fakdi, Joaquim Mel

endez and Roberto Petite

Control Engineering and Intelligent Systems Group eXIT, University of Girona

Campus Montilivi Building PIV, 17071 Girona, Spain

Keywords:

Clinical decision support systems, Case-based reasoning, TAVI, Clinical workﬂow.

Abstract:

A Clinical Decision Support System (CDSS) is an interactive software designed to assist physicians, and

other health professionals, with health decision-making tasks through an inference process that associates

observations with conclusions supported by objective and expert knowledge on a speciﬁc application domain.

As complexity of surgical processes arises, there is a demand from the clinicians to extend the capabilities

from a pure research database towards a clinically integrated decision support system. The work presented

in this paper details a proposal of a CDSS designed to to assess clinicians during interventions by retrieving

and reusing past similar cases. Following an operative workﬂow, the designed software will use a Case-Based

Reasoning (CBR) methodology to retrieve similar past cases from a case base to give case-speciﬁc advice

at any particular point of the process, specially during preoperative and intraoperative stages. The proposal

intends to develop a monitoring mechanism capable to recognize relevant data to every stage in the intervention

workﬂow. Thus, measures and observations generated during the intervention will be used to retrieve the most

similar cases occurred in previous occasions. The paper proposes an example application of the CBR strategy,

where it is adapted to a recent complex minimally invasive intervention which is receiving growing attention

lately, the Transcatheter Aortic Valve Implantation (TAVI).

1 INTRODUCTION

In a general term, CDSSs comprise a large spectrum

of systems which provide clinicians, staff, patients,

and other individuals with knowledge and person-

speciﬁc information, intelligently ﬁltered and pre-

sented at appropriate times, to enhance health and

health care (Berner, 2009). CDSSs have been used

in clinical practice since 1970, but despite several

individual success since then, their impact on rou-

tine clinical practice has not been as strong as ex-

pected, specially due to the barriers to their im-

plementation, which still remain in place. CDSSs

should form an important part of the ﬁeld of clin-

ical knowledge management technologies through

their capacity to support the clinical process and use

of knowledge,including knowledge maintenance and

continuous learning, from diagnosis and investigation

through surgery, treatment and long-term care. Argu-

ments for and against the value of CDSSs have been

discussed over the years. Among its potential ben-

eﬁts, as stated in (Coiera, 2003), CDSSs could im-

prove patient safety through reduced medication er-

rors and adverse events. Also, they should improve

quality of care by increasing pathways, guidelines and

documentation available for patients. Another advan-

tage is that CDSSs may improve efﬁciency in health

care delivery, reducing costs through faster order pro-

cessing and avoiding test duplication. As drawbacks,

clinicians may see CDSSs as a threat to clinical judg-

ment and sometimes too inﬂexible, with difﬁculties

to depart from ordered, pre-prepared paths. Also,

computer-supported decision systems promote over-

reliance on software decisions which may limit clin-

icians’ freedom to think at some point. In the same

way, bad designed systems can create extra work or

extend clinical procedures more than necessary. Fi-

nally, maintenance costs, professional support and

training needed by the medical staff in order to use

the software properly could also be seen as arguments

against the utilization of CDSSs.

Decision support can be provided at various stages

in the care process, from preventive care through di-

agnosis and treatment to monitoring and follow-up.

As detailed in (Perreault and Metzger, 1999), com-

puter CDSSs can be designed to support four ba-

201

El-Fakdi A., Meléndez J. and Petite R..

A PROPOSAL OF A WORKFLOW-BASED CBR FOR INTERVENTIONAL CDSS - Application to Transcatheter Aortic Valve Implantation (TAVI).

DOI: 10.5220/0003771302010206

In Proceedings of the International Conference on Health Informatics (HEALTHINF-2012), pages 201-206

ISBN: 978-989-8425-88-1

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

sic clinical functions. First, giving administrative

support, aiding in clinical coding and documenta-

tion, authorization of procedures, and referrals. Sec-

ond, managing clinical complexity, keeping patients

on research and chemotherapy protocols, tracking or-

ders, referrals follow-up, and preventive care. The

third function deals with cost control, programming

CDSSs to monitor medication orders with the objec-

tive of avoiding duplicate or unnecessary tests. A

fourth stage of application, which represents the fo-

cus of this work, involves CDSSs with low level de-

cision support, helping in clinical diagnosis and treat-

ment plan processes, giving case-speciﬁc support in

highly complex surgery operations or Minimally In-

vasive Surgerys (MISs), promoting use of best prac-

tices and guidelines based on population case man-

agement.

The work presented in this paper details a proposal

of a CDSS designed to give case-speciﬁc assessment

to clinicians during complex surgery or minimally in-

vasive interventions. Following a perioperative work-

ﬂow, the designed software will use a Case-Based

Reasoning (CBR) methodology (see Figure 2) to re-

trieve similar past cases from a case base to give case-

speciﬁc advice at any particular point of the process,

specially during preoperative and intraoperative steps.

This support will be presented in the form of 2D/3D

images, numerical data or decisions taken based on

past similar cases. The intervention will be moni-

torized and used to recognize similar previous cases

at any time during the intervention. The paper pro-

poses an example application of the method, adapted

to assist a recent complex minimally invasive surgery

which is receiving growing attention lately, the Tran-

scatheter Aortic Valve Implantation (TAVI). This pa-

per is structured as follows. In Section 2 recent appli-

cations of CDSS for surgical processes are reviewed.

Also, a description of the workﬂow management dur-

ing an intervention is detailed. Section 3 describes

the proposed workﬂow-based CBR for assessment in

complex surgical procedures. In Section 4, the pro-

posed support system is exempliﬁed with its applica-

tion to a minimally invasive surgical procedure, TAVI.

Finally, conclusions and the future work to be done

are included in Section 5.

2 CDSS INTEGRATION WITH

CLINICIAN WORKFLOW

Over the past few years, signiﬁcant research in the

area of medical informatics points to the importance

of understanding workﬂow processes to support the

development of CDSSs for complex workspaces (Sit-

tig et al., 2008; Patel et al., 2001). Nowadays there is a

demand from the clinicians to extend the capabilities

from a pure research database towards a clinically in-

tegrated decision support system. Recent introduction

of new clinical techniques such as Minimally Invasive

Surgery (MIS) has led to several technological inno-

vations inside the Operation Room (OR). All these

advances, however, create new difﬁculties, such as

inadequate information transparency, limited access,

and poor visualization. Therefore, clinicians must

rely on advancements in medical imaging technology

(Dugas et al., 2002). These limitations in MIS are

constantly giving rise to new research and develop-

ment in the area of decision support systems. Such

systems are providing realtime image guidance and

task automation support while the clinician is per-

forming the intra-operative tasks (Wood et al., 2007).

Empirical studies demonstrate the beneﬁts of includ-

ing decision support into complex workspace scenar-

ios, which lead to safer working environments and

prevention of errors (Durieux et al., 2000). Some

examples of successful applications of CDSSs into

clinical workﬂows comprise computer based patient

record systems (Patel et al., 2000), knowledge man-

agement systems for biomedical engineering (Rinkus

et al., 2004) and computer based training systems in

pathology (Crowley et al., 2003). Although multiple

factors are believed to affect the success or failure of

CDSS intervention implementation, a helpful CDSS

is dependent on the completeness and accuracy of the

evidence base used to support it. Also, from the suc-

cessful applications mentioned before it can be ex-

tracted that integration with workﬂow is key to suc-

cess. How to integrate the CDSS with clinician work-

ﬂow, however, remains a challenge, in part because

there are no current standards for clinical workﬂow

(Das and Eichner, 2010).

In order to implement a workﬂow-based CDSS,

timing and involved subprocesses must by identiﬁed

ﬁrst. Figure 1 shows a schematic workﬂow of an ex-

empliﬁed operative process. Although the main in-

formation and support will be given during the intra-

operative phase, some important aspects belonging

to the pre and post-operative phases can also be in-

cluded. As Figure 1 shows, the whole procedure has

been split into tasks, a task being any particular step

taken e.g. apply anesthesia, initial puncture location

or valve ﬁnal placement. Each task has a different

number of distinguishable items or attributes associ-

ated. These attributes can be described as numerical

data, text data or boolean data. During the interven-

tion execution, and according to the information be-

ing generated, the CDSS has to be capable to identify

current task in the workﬂow and search similar cases

HEALTHINF 2012 - International Conference on Health Informatics

202

Figure 1: Schematic workﬂow showing an exempliﬁed operative process.

in the case base.

The main idea of the proposed computer CDSS

is to deﬁne a case structure for every intervention

based on the operation workﬂow we want to give sup-

port to. With a case base of indexed past interven-

tions (organized as cases), clinicians would be able

to retrieve on-line any similar case, task or attribute

at any point along the surgical intervention together

with a formed new solution, by adapting/combining

solutions of the retrieved cases, for each of the simi-

lar cases which may help them to select the next step

properly. Also, new, revised cases will be added to the

case base for future use. The designed software will

use a CBR methodology and to give case-speciﬁc ad-

vice at any particular point of the intervention work-

ﬂow, specially focused along preoperative and intra-

operative phases.

3 WORKFLOW-BASED CBR FOR

DECISION SUPPORT

CBR is a technique of artiﬁcial intelligence that at-

tempts to solve a given problem within a speciﬁc

domain by adapting established solutions to similar

problems (Aamodt and Plaza, 1994). As can be seen

in Figure 2, CBR has been formalized for purposes

of reasoning and learning based on the exploitation of

existing similar historical records as humans do. A

basic four-steps procedure is deﬁned: retrieve, reuse,

revise and retain. Given a target problem, during the

retrieve step, cases from memory that are relevant to

solving it are recovered. A case is a whole although

usually two parts can be identiﬁed according to the

problem to be solved: problem and solution space.

Problem space can incorporate speciﬁc case informa-

Figure 2: Four-step process of Case-Based Reasoning.

tion and also contextual information useful for solv-

ing the problem whereas solution space incorporates

information directly related with the solution for solv-

ing the problem. While in the reuse phase, solutions

from previous cases are mapped to target the current

problem. This may involve adapting the solution as

needed to ﬁt the new situation. Having mapped the

previous solution to the target situation, during the re-

vise step the new solution is tested and, if necessary,

checked by clinicians. After the solution has been

successfully adapted to the target problem, the retain

last step concerns with storing the resulting experi-

ence as a new case in the case base. These features

make CBR a good contender for any decision support

system (Watson, 2002). CBR has been used in other

medical decision support systems. An integration of

CBR and rule-based reasoning was used in systems

for the planning of ongoing care of Alzheimer’s pa-

tients (Marling and Whitehouse, 2001) and for the

management of Diabetes patients (Bellazi et al., 1999)

In order to design a case-based system adapted

to an operation workﬂow like the one described in

Section 2, we should determine a case model, case

A PROPOSAL OF A WORKFLOW-BASED CBR FOR INTERVENTIONAL CDSS - Application to Transcatheter Aortic

Valve Implantation (TAVI)

203

Figure 3: Case structure and retrieval process along the op-

erative workﬂow.

indexing and a similarity metric. As illustrated in

Figure 3, every case of the case base is deﬁned as a

set of attributes classiﬁed among the different tasks

which compound the whole operative workﬂow. The

stored cases do not need to contain the same number

of attributes, as it will depend on the amount of infor-

mation available when storing each case. Therefore,

some cases will be richly ﬁlled with useful informa-

tion concerning all the tasks of the workﬂow while

others can be poorly deﬁned with missing informa-

tion in certain tasks or even non-existent tasks at all,

as not all of them are mandatory and depend on each

patient.

When the intervention begins, parallel to the

workﬂow progression described in Figure 3, the CBR

mechanism retrieves speciﬁc and contextual problem

information of similar past cases. Such assessment

step determines which cases address problems most

similar to the current problem, to identify them as

starting points for solving the new problem. Next,

the CBR gives the clinicians information about the

solution for solving the problem by case adaptation,

which forms a new solution by adapting/combining

solutions of the retrieved problems. Once the oper-

ation ends, clinicians proceed with the new case re-

vision, which evaluates and adjusts the adapted solu-

tions and, if suitable, the new case will be retained

with the system learning from the situation by storing

the result as a new case for future use.

4 EXAMPLE OF APPLICATION:

TRANSCATHETER VALVE

IMPLANTATION

In Transcatheter Aortic Valve Implantation (TAVI)

(Webb and Cribier, 2011), a synthetic valve is trans-

ported to the heart through a small hole made in groin.

This procedure can be compared to that performed

when placing a stent, or performing balloon angio-

plasty. This technique was ﬁrst developed in Europe,

where it was initially performed in 2002. Since then,

more than 10000 patients have beneﬁt from it and the

results have shown the procedure to be effective in im-

proving functioning in the patients with severe aortic

stenosis. In the recent years TAVI is assuming a major

role in the routine management of patients with aortic

stenosis and now TAVI is considered the standard in

patients who are not candidates for conventional sur-

gical Aortic Valve Replacement (AVR). On the basis

of almost 10 years of experience TAVI also appears

to be a reasonable option for some operable, but high-

risk patients. Nevertheless considerable work needs

to be done before TAVI is expanded into lower risk

groups.

Figure 4 shows the speciﬁc workﬂow for the TAVI

example. As can be appreciated, the procedure (from

pre-operative to post-operative treatment) is subdi-

vided in 8 different tasks, starting with a study of pa-

tient suitability for TAVI in Task 0, to Task 7 where

the patient is ﬁnally transferred to the Intensive Care

Unit (ICU). As illustrated in Figure 4, the work-

ﬂow does not evolve only in a forward direction but

it also turns backward if necessary, specially along

the preoperative phase, where some contraindications

could arise during the study which may turn the pa-

tient not eligible for TAVI. Parallel to the workﬂow

progress, as can be seen along the bottom section of

Figure 4, all the tasks are linked with the different in-

formation sources which will provide the necessary

attribute data required for deﬁning the patient case

model. For the particular case of a MIS like TAVI,

most of the information is acquired through imaging

system devices.

As shown in Table 1, every attribute of the case

model is matched to a task or group of tasks where

clinicians consider it to be relevant or critic for the

correct development of the operative process. Also,

the attributes are linked with the source images or

monitoring sensors used to assign them a value. As

sources of different accuracy can be used to obtain

the same attribute, the information will always be pre-

sented together with its origin so the clinicians have

the last word to decide which source data is more re-

liable.

HEALTHINF 2012 - International Conference on Health Informatics

204

Figure 4: TAVI workﬂow showing tasks and data source where attributes are extracted from.

Table 1: Attributes obtained from source data and linked to the workﬂow tasks.

Attribute Name Task Number Data Source

Euroscore Task 0 -

Mitral regurgitation Task 0 US, A, CT, MRI

Other pathologies Task 0 US, A, CT, MRI

Vascular diameters Task 1, Task 3 CT, US, F, MRI

Calciﬁcations Task 1, Task 3 US, F, CT, MRI

Tortuosity Task 1, Task 3 US, F, CT, MRI

Atherosclerosis Task 1, Task 3 US, F, CT, MRI

Porcelain aorta Task 1, Task 3 US, F, CT, MRI

Uncoiled aorta Task 1, Task 3 US, F, CT, MRI

Iliac stenosis Task 1 US, F, CT, MRI

Annular diameter Task 1 US, A, CT, MRI

Sinus of Valsalva diameter Task 1 US, A, CT, MRI

Ascending aorta diameter Task 1 US, A, CT, MRI

Height of coronary arteries Task 1 US, A, CT, MRI

Blood Count Task 2, Task 3, Task 4, Task 5, Task 6, Task 7 Sensor

Blood Pressure Task 2, Task 3, Task 4, Task 5, Task 6, Task 7 Sensor

Coagulation Task 2, Task 3, Task 4, Task 5, Task 6, Task 7 Sensor

Electrolytes Task 2, Task 3, Task 4, Task 5, Task 6, Task 7 Sensor

Vessel Rocked Task 3 US, F, A

Cardiac pacing value Task 4, Task 5 US, A

Inﬂate volume 1 Task 4 US, A

Inﬂate volume 2 Task 5 US, A

Valve planes Task 5, Task 6 US, A, CT

Valve leak Task 6, Task 7 US, A

Coronary ﬂow Task 6, Task 7 US, A

Aortic/Iliac injury Task 6, Task 7 US, F

Femoral ﬂow Task 6, Task 7 US, F

Valve completely open Task 6, Task 7 US, F

5 CONCLUSIONS AND FUTURE

WORK

The work presented in this paper details a proposal of

a CDSS designed to give case-speciﬁc assessment to

clinicians during complex surgery or minimally inva-

sive interventions.

The work progress is solid and consistent and, up

to this point, the workﬂow has been structured and

a case model deﬁned together with specialized clini-

A PROPOSAL OF A WORKFLOW-BASED CBR FOR INTERVENTIONAL CDSS - Application to Transcatheter Aortic

Valve Implantation (TAVI)

205

cians. Therefore, our efforts are currently centered on

building an initial case base. For this purpose, past

cases from collaborating hospitals and medical cen-

ters have been collected and they are being translated

to match the case model previously deﬁned. First sim-

ulated results are expected to be obtained soon be-

fore the end of the year. Also, more medical centers

have been contacted in order to spread the case base

along different hospitals within a cluster, and thus al-

lowing clinicians from different teams to consult case

problems and solutions from colleagues of other in-

stitutions. To ﬁnish, it is important to remark that the

feedback received from the numerous physicians con-

sulted has been very positive and the interest showed

in this project is encouraging. Clinicians agree that

CDSS are of high value specially in recently grow-

ing MIS like TAVI where the number of interventions

per year is still low compared to common surgery and

which also require expert hands due to its complex

procedure.

ACKNOWLEDGEMENTS

We would like to give our special thanks to the Hos-

pital Clinic of Barcelona for helping us with the case

model deﬁnition and giving us example cases of TAVI

patients. This work has been ﬁnanced by the Span-

ish Government Commission Ministerio de Indus-

tria, Turismo y Comercio (MITyC) under the project

PLAN AVANZA 2 labeled by Information Technol-

ogy for European Advancement 2 (ITEA2).

REFERENCES

Aamodt, A. and Plaza, E. (1994). Case-based reasoning:

Foundational issues, methodological variations, and

system approaches. Artiﬁcial Intelligence Communi-

cations, 7:39–52.

Bellazi, R., Montani, S., Portinale, L., and Riva, A. (1999).

Integrating rule-based and case- based decision mak-

ing in diabetic patient management. In 3rd Inter-

national Conference in Case-Based Reasoning IC-

CBR’99, pages 386–400, Seeon Monastery, Bavaria,

Germany.

Berner, E. (2009). Clinical decision support systems:

State of the art. AHQR publication no. 09-0069-EF,

Agency for Healthcare Research and Quality.

Coiera, E. (October 2003). The Guide to Health Informatics

(2nd Edition). Arnold, London, UK.

Crowley, R., Naus, G., and Friedman, C. (2003). Develop-

ment of visual diagnostic expertise in pathology: An

information processing study. Journal of American

Medical Informatics Association, 10(1):39–51.

Das, M. and Eichner, J. (2010). Challenges and barriers

to clinical decision support design and implementa-

tion experienced in the agency for healthcare research

and quality cds demonstrations. AHQR publication

no. 10-0064-EF, Agency for Healthcare Research and

Quality.

Dugas, M., Schauer, R., Volk, A., and Rau, H. (2002). In-

teractive decision support in hepatic surgery. BMC

Medical Informatics and Decision Making, 2.

Durieux, P., Nizard, R., Ravaud, P., Mounier, N., and Lep-

age, E. (2000). A clinical decision support system

for prevention of venous thromboembolism: effect on

physician behavior. Journal of the American Medical

Association, 283(21):2816–21.

Marling, C. and Whitehouse, P. (2001). Case-based rea-

soning in the care of alzheimer’s disease patients. In

4th International Conference in Case-Based Reason-

ing ICCBR’01, pages 702–715, Vancouver, Canada.

Patel, V., Arocha, J., and Kaufman, D. (2001). Primer on

aspects of cognition for medical informatics. Journal

of American Medical Informatics Association, 8:324–

343.

Patel, V., Kushniruk, A., Yang, S., and Yale, J. (2000). Im-

pact of a computer-based patient record system on

data collection,knowledge organisation, and reason-

ing. Journal of American Medical Informatics Asso-

ciation, 7(6):569–85.

Perreault, L. and Metzger, J. (1999). A pragmatic frame-

work for understanding clinical decision support.

Journal of the Healthcare Information and Manage-

ment Systems Society, 13(2):5–21.

Rinkus, S., Walji, M., Johnson-Throop, K., Malin, J., Tur-

ley, J., and Smith, J. (2004). Human-centerd design of

a distributed knowledge management system. Journal

of Biomedical Informatics, 38:4–17.

Sittig, D., Wright, A., Osheroff, J., Middleton, B., Teich, J.,

Ash, J., Campbell, E., and Bates, D. (2008). Grand

challenges in clinical decision support. Journal of

Biomedical Informatics, 41(2):387–392.

Watson, I. (2002). Applying knowledge management:

Techniques for building organisational memories. In

6th European Conference in Advances in Case-Based

Reasoning ECCBR’02, pages 6–12, Aberdeen, Scot-

land, UK.

Webb, J. and Cribier, A. (2011). Percutaneous transarterial

aortic valve implantation: what do we know? Euro-

pean Heart Journal, 32:140–147.

Wood, B., Locklin, R., Viswanathan, A., Kruecker, J.,

Haemmerich, D., and Cebral, J. (2007). Technolo-

gies for guidance of radiology ablation in the multi-

modality interventional suite of the future. Journal of

Vascular Interventional Radiology, 18:9–24.

HEALTHINF 2012 - International Conference on Health Informatics

206