Towards an “Operational” Educational Model in Healthcare:

Exploiting Computer-Interpretable Guidelines

Alessio Bottrighi

, Gianpaolo Molino

, Luca Piovesan

and Paolo Terenziani

DISIT, Università del Piemonte Orientale, Viale Teresa Michel 11, Alessandria, Italy

Az. Osp. San Giovanni Battista, C.sa Bramante 88, Torino, Italy

Keywords: Clinical Guidelines, Practitioners, Healthcare-agents, Medical Students Education.

Abstract: Clinical guidelines (GLs) encode the best medical practices. GLs have been widely exploited to enhance the

quality of patient care, and to optimize it, and several computer-based approaches to manage computer-

interpretable guidelines (CIGs) have been proposed in the literature. Quite surprisingly, however, the

potentialities of CIG systems in medical education have not been considered yet. In this position paper we

argue that, since CIG systems support the “simulation” of the application of GLs on specific patients, they

can be used to show students how to apply medical knowledge and best practices on specific patients.

Therefore, using CIG systems, students may learn an “operational methodology” that, otherwise, they could

only learn from the medical practice. In this paper, we have taken GLARE (and its extension, META-GLARE)

as an example of CIG system, and we have addressed the roadmap we intend to follow to fully exploit its

potentialities in medical education.

1 INTRODUCTION

Clinical guidelines (GLs) have been defined as

“systematically developed statements to assist

practitioner and patient decisions about appropriate

healthcare under specific clinical circumstances”

(Field and Lohr, 1990). They encode best medical

practices, promoting the adoption of evidence-based

medicine and supporting the quality and the

standardization of healthcare services, and the

optimization of costs. Thousands of GLs have been

devised. For example, the Guideline International

Network (http://www.g-i-n.net) groups 106

organisations in 54 countries, and provides a library

of more than 6500 GLs. In this paper, we argue that

such a valuable body of knowledge may be paired

with software tools supporting its application and

simulation (the so-called Computer-Interpretable

Guideline Systems, see below), and fruitably used to

complement “traditional” education in medicine.

Despite the large number of GLs, and their

diffusion, GLs have not fully achieved all their goals,

in terms of quality and optimization of the healthcare

services. The discussion of the reasons for GL not-

full success is outside the goals of this paper. Here we

just highlight two of such reasons, which have largely

motivated the introduction of ICT support tools to GL

management. Non-computerized GLs are large

bodies of knowledge (even hundreds of pages),

mostly expressed as free text, describing the “best

practice” recommendations for the treatment of a

given disease. Using such large bodies of (textual)

knowledge to diagnose and treat a specific patient is

a difficult issue for physicians, who are left alone

(i) to interpret the textual description (which, as

any natural language text, contains

imprecisions and ambiguities)

(ii) to identify the “mapping” between the general

recommendations in the GL and the specific

patient (and disease or clinical condition) at

hand; indeed, when diagnosing or treating a

patient, physicians should quickly identify,

among pages and pages of free text, the few

parts that are relevant for the specific patient

at hand.

Considering (among the others) issues (i) and (ii)

above, since the 90’s, the medical community has

started to develop many different systems and

projects to support physicians in the management of

GLs. In particular, in Computer-Interpretable

Guidelines (CIG) systems, GLs

(1) are represented in a formal and unambiguous

way, and

402

Bottrighi, A., Molino, G., Piovesan, L. and Terenziani, P.

Towards an “Operational” Educational Model in Healthcare: Exploiting Computer-Interpretable Guidelines.

DOI: 10.5220/0007482604020409

In Proceedings of the 12th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2019), pages 402-409

ISBN: 978-989-758-353-7

(2) most CIG systems provide execution tools that

are automatically connected to the patients’

clinical records, so that they are able to

automatically detect the GL recommendations

that are more appropriate for the given patient.

Besides disease-based support tools (e.g., software

tools considering only a specific disease), several

disease-independent tools for CIGs have been

proposed in the literature (see e.g. (Fridsma, 2001;

Gordon and Christensen, 1995; Peleg, 2013)). Such

systems usually provide facilities to acquire,

represent and/or execute GLs in different (medical)

domains. Such tools have been mainly developed to

support physicians in patient care, while, quite

surprisingly, their impact in medical education has

been quite neglected until now.

2 CIG SYSTEMS AND

EDUCATION

Since GLs encode “best medical practices”, their role

in the medical practice has been widely investigated.

Indeed, several studies have shown that the quality of

patient treatments is higher in case physicians have

been educated to adopt GLs (see, e.g., (Corriere et al.,

2014) as regards the case of diabetes).

CIG systems have been quite widely used in order

to acquire GLs into a computer format, and to support

practitioners in the diagnosis and treatment of patients.

However, to the best of our knowledge, the

possible impact of the adoption of CIG systems in

education has not been investigated yet. Indeed, the

main claim of this position paper is that although CIG

systems until now have been conceived as a support

to practitioners to deal with patients affected by

specific diseases, they can evolve to support also

medical education. Indeed, in this paper we argue that

CIGs have the potentiality of drastically improving

the education of healthcare agents (medical

students/ practitioners/ nurses), proposing them an

“operational” methodology showing how to (reason

in order to) apply best practice recommendations to

specific patients.

First of all, it is important to point out that the

approach we propose may support the education of

both practitioners (e.g., their continuous education)

and medical students. Indeed, though the former have

responsibilities that the latter do not have, both of

them have to learn how to apply best medical

practices to patients. Indeed, for the same reason, also

other types of healthcare agents (e.g., nurses) can take

advantage of our approach.

Second, we want to highlight that the approach we

propose is not intended to substitute “traditional”

education in medicine (erogated through courses and

textbooks), but to complement it.

As a matter of facts, “traditional” education

provided by medical texts and courses covers a wide

range of knowledge, ranging from human anatomy to

the description of diseases, and of their treatments.

While such aspects are certainly very important (and

necessarily needed), another aspect should also

deserve a specific attention, the “operational” (our

terminology) aspect: how to “operate” on a specific

patient? How to proceed to diagnose and treat

her\him through the best medical practices? Such a

kind of “operational” knowledge is usually not

considered in textbooks, so that it can be learned by

medical students (and healthcare agents in general)

only “by practicing”.

On the other hand, CIG systems have the potentiality

of enriching education through an operational

methodology, showing how to proceed to diagnose

and to treat specific patients. Indeed, as discussed in

the introduction, any CIG supports healthcare agents

in the treatment of a specific patient, by automatically

“focusing” on the part of the GL which is relevant to

the (status of) the patient at hand (see problem (ii) and

the solution (2) in the introduction). This is, indeed,

the process that students have to learn to be able to

cope with real patients: how to focus on the more

appropriate parts of their general medical knowledge

and of the best practices and to apply them when

considering the patient.

Until now, CIG system have been developed in

order to support practitioners in the diagnose and

treatment of patients. To do so, most CIG systems

provide execution modules, which support

physicians in the application of a general GL to a

specific patient. Execution modules take in input the

patient’s clinical record and, on the basis of the

patient’s data, suggest the proper actions and, above

all, they help physicians to take the decision

appropriate to the patient. Intuitively speaking, one

can say that CIG execution tools support the focusing

on the specific parts of the general GL that are

appropriate for the patient at hand, and help to take

the best decisions on the basis of the patient’s data. In

other words, they make best medical practices

“operational”, by supporting the application of

general GLs to the patients.

In this paper, we propose a new use of CIG

systems. Instead of being used to support physicians

in the execution (application) of a GL to a real patient,

they could also be used in education, to simulate such

an execution (application). The learning healthcare

Towards an “Operational” Educational Model in Healthcare: Exploiting Computer-Interpretable Guidelines

403

Figure 1: Architecture of GLARE. Rectangles represent computation modules, and ovals data/knowledge bases.

agent (e.g., student, practitioner) is given a patient

(in the form of the history of the evolution of the

patient’s data), and the CIG system can be used to

show how the GL would recommend to operate on

her\him (while, of course, no action is “physically”

executed: it is just a simulation).

Notably, since it is a simulation, the presence of

“physical” patients is not required: all that is needed

is the evolution of patients’ clinical data. As a

remarkable consequence, teachers may propose to

students (the data of) “significant” patients, and\or

invent (the data of) patients in such a way to force

students to explore the diagnosis/treatment of most

important (and\or problematic) cases.

This contribution may provide, in our opinion,

crucial benefits in the area of education, to

complement textbooks and traditional courses with

“practical” examples (through simulation) of how to

apply medical knowledge on specific patients.

Indeed, different modalities of education through

simulations can be provided:

(i) “standard” simulation, in which the CIG

system is used to show students how the (real

or invented) patient should be treated, step-by-

step, given the CIG recommendations

(ii) “second-opinion” simulation, in which a

student has to indicate how s\he would treat a

(real or invented) patient, and the CIG system

is used to indicate to the student where s\he has

followed the recommendations of the CIG,

and where s\he has violated them (with

additional explanations).

Notably, the educational approach we propose can be

strongly based on the facilities provided by currently

existing CIG systems. In particular, the acquisition of

CIGs can be done through the acquisition tools

already provided by such systems, with usually

automatically interact with patient clinical records.

As a consequence, the starting point of each

simulation could simply be the loading of

- an already acquired CIG

- the evolution of the data of a (real or

invented) patient.

As regards the simulation, “standard” simulation can

be performed by taking advantage of the execution

module already provided by most CIG systems. On

the other hand, the “second-opinion” simulation

require a modification of current execution tools, as

discussed in the next Section.

While in Section 4 we further elaborate such

issues, here we highlight that, for education, the

disease-independent CIG systems have a major

advantage with respect to the disease-based ones. In

fact, with such systems, one can acquire a library of

different CIGs (expressed using the same formalism,

the one provided by the CIG system), and provide a

unique software tool for simulating each of them.

Therefore, they allow one to develop a uniform

educational environment, in which all the CIGs are

homogeneously represented through the same

formalism, and the same simulation mechanism is

applied to all of them. On the contrary, the adoption

of disease-dependent CIG systems would force the

adoption of multiple different representation

formalisms and simulation mechanisms, one for each

different CIG. This move would force students to

learn different formalisms and to practice different

software tools for simulation, which would only be a

loss of time for medical students.

Though our position and the discussion until

now is fully general, in our future research we aim at

making it fully operative taking advantage of GLARE

(Guideline Acquisition, Representation and

Execution; (Terenziani et al., 2008)), and its

extension, META-GLARE (Bottrighi and Terenziani,

2016). Before highlighting how we plan to cope with

education through CIGs, in Section 3 we quickly

mention the current status of GLARE and META-

GLARE.

HEALTHINF 2019 - 12th International Conference on Health Informatics

404

3 STATUS OF GLARE PROJECT

GLARE is a disease-independent system for the

acquisition and execution of CIGs, which we are

developing since 1997, in collaboration with the

physicians of Azienda Ospedaliera San Giovanni

Battista in Torino, Italy (one of the major hospitals in

Italy). The core of GLARE (see the reference and the

box on the left of Fig. 1) is a modular architecture.

CG_KRM (Clinical Guidelines Knowledge

Representation Manager) is the main module of the

system: it manages the internal representation of GLs,

and operates as a domain-independent and task-

independent knowledge server for the other modules;

moreover, it permanently stores the acquired GLs in

a dedicated Clinical Guidelines Database (CG-DB).

The Clinical Guidelines Acquisition Manager

(CG_AM) provides expert-physicians with a user-

friendly graphical interface to introduce the GL into

the CG_KRM and to describe them. It may interact

with four databases: the Pharmacological DB, storing

a structured list of drugs and their costs; the

Resources DB, listing the resources that are available

in a given hospital; the ICD DB, containing an

international coding system of diseases; the Clinical

DB, providing a “standard” terminology to be used

when building a new GL, and storing the descriptions

and the set of possible values of clinical findings. The

execution module (CG-EM) executes a GL for a

specific patient, considering the patient’s data

(retrieved from the Patient DB). CG-EM stores the

execution status in another DB (CG Instances) and

interacts with the user-physician via a graphical

interface (CG-IM).

GLARE’s architecture is open. In the latest years,

several new modules and\or methodologies have been

added to cope with automatic resource-based

contextualization (ADAPT module, (Terenziani et

al., 2004)), temporal reasoning (TR, (Anselma et al.,

2006)), decision making support (DECIDE_HELP,

(Montani et al., 2005)), model-based verification

(VERIFY, (Bottrighi et al., 2010)), and comorbidities

(COMORBID, (Piovesan et al., 2014) (Piovesan et

al., 2015).

Representation Formalism. The core of GLARE

is the definition of the representation formalism used

to model GLs. Notably, a unique (disease-

independent) formalism is provided, and is used by

all the modules of GLARE. In GLARE, a GL is

represented through the set of actions composing it.

GLARE distinguishes between atomic and composite

actions. Atomic actions can be regarded as

elementary steps in a GL, in the sense that they do not

need a further decomposition into sub-actions to be

executed. Composite actions are defined in terms of

their components (atomic or composite actions), via

the “has-part” relation. GLARE adopts four different

types of atomic actions. Work actions represent basic

atomic actions which must be executed on the patient,

and can be described in terms of a set of attributes,

such as name, (textual) description, cost, time,

resources, goals. Query actions are requests of

information, which can be obtained from the outside

world (physicians, Databases, knowledge bases).

Conclusions represent the output of decision actions.

Decision actions are specific types of actions

embodying the criteria which can be used to select

alternative paths in a GL. They are crucial also to the

education task, so that they are detailed in the

following subsection.

Control relations establish which actions might

be executed next and in what order. We distinguish

among four different control relations: sequence,

constrained, alternative and repetition. Temporal

constraints between actions (e.g., overlaps, during)

are also supported.

Testing. GLARE has been applied to different

medical domains, including bladder cancer, reflux

esophagitis, heart failure, and ischemic stroke.

META-GLARE

In the last years, we have defined a new GL system,

META-GLARE (Bottrighi and Terenziani, 2016), on

top of GLARE. META-GLARE has been designed

to make extensions of CIG formalisms easy to

implement-manage, so that its availability is very

important, to support the extensions of GLARE with

the constructs needed to support educations.

However, META-GLARE is a support for system

designers, while users (e.g., physicians, or students)

never have to directly interact with META-GLARE.

3.1 Decisions in

GLARE\META-GLARE

Decisions are probably the most crucial aspect of

GLs, since they allow user-physicians to

discriminate among alternative actions. GLARE

supports two different types of decisions: diagnostic

and therapeutic decisions.

Diagnostic decisions consider a set of

parameters (to be evaluated on the basis of the status

of the patient) to discriminate among different

diagnoses. Of course, different parameters have to be

considered, depending on the specific diagnostic

decision. In GLARE, we support “scored” diagnostic

decisions: for each one of the relevant parameter (e.g.

Towards an “Operational” Educational Model in Healthcare: Exploiting Computer-Interpretable Guidelines

405

Figure 2: Tabular representation of a diagnostic decision for GERD, as shown in GLARE acquisition module.

Figure 3: Tabular representation of the therapeutic decision for “possible GERD”, as shown in GLARE acquisition module.

fever) and considering the alternatives to be

discriminated, expert-physicians define a priori a set

of values/ranges (e.g., fever < 37, between 37 and 38,

between 38 and 39, > 39; or, qualitatively: no, low,

medium, high). For each one of the values/ranges of

the findings in each one of the alternatives, the

expert-physician defines a score. Moreover, a

threshold is fixed to separate recommended/disliked

actions. GLARE execution engine considers the

patient’s data to evaluate each parameter, and, for

each alternative diagnosis, sums up the scores. Only

the alternatives whose additive score is greater than

the threshold are recommended for selection to the

user. To summarize, a decision among a set of

diagnoses can be represented as an open set of triples

<diagnosis, parameter, score> (where, in turn, a

parameter is a triple <data, attribute, value>), plus a

threshold which is relative to the sum of the scores.

For example, Fig. 2 shows a diagnostic decision

within the Gastro-Esophageal Reflux Disease

(GERD) guideline, namely the “differential

diagnosis”, which allows the physician to

discriminate between “possible GERD”, “probable

GERD” and “no GERD” according to the values of

several parameters: heartburn absent (abbreviated by

“no-hb” in Fig. 2), or lasted not more than 3 months

(“hb≤3”), or lasted more than 3 months (“hb>3m”);

dysphagia absent (“no-dys”) or present (“dys”);

occurrence of weight loss (“wl”) or non-occurrence

(“no-wl”); absence (“no-hem”) or presence (“hem”)

of hematemesis; postural reflux absent (“no-ref”), or

lasted not more than 3 months (“ref≤3”) or more than

three months (“ref>3m”). The threshold for such a

decision (not shown in Fig. 2) is >9. E.g., given the

scores in Fig. 2, one should conclude “no GERD”

only if heartburn, dysphagia, weight loss,

hematemesis and postural reflux are all absent.

Notably, also Boolean diagnostic decisions are

supported by META-GLARE, and will be used in the

educational task.

On the other hand, in the therapeutic decisions,

physicians have to choose from different therapies

(treatments) according to a given set of parameters:

effectiveness, cost, side-effects, compliance,

duration, which have to be specified for each one of

the alternative therapies to be discriminated. Thus, a

decision action can be represented by the set of the

parameters above, for each one of the alternatives.

E.g., in Fig. 3 we show a possible way of

modelling the decision among four different

treatments in the case of “possible GERD”, in the

tabular fashion which is used by GLARE acquisition

module.

3.2 GLARE in Education: First Steps

GLARE has been already tested in the training of

physicians in the context of emergency medicine, to

cope with poly-trauma, within the ROPHS (Report on

the Piedmont Health System) project (Leonardi et al.,

2012). In such a project, GLARE has been used “as it

is” for educational purposes following the lines

suggested in Section 2, point (1). Specifically:

(i) together with expert physicians, we have

acquired a GL for polytrauma in GLARE

(ii) expert physicians have defined for us a set of

“typical” patients affected by polytrauma. We

HEALTHINF 2019 - 12th International Conference on Health Informatics

406

have stored the evolution of the data of such

patients in the DB of clinical records.

(iii) Expert physicians have used GLARE execution

module to provide a “standard” simulation (see

Section 2) of the treatment of the patients

The adoption of GLARE has supported the

physicians attending the course while learning “the

process”: we showed the simulation of the application

of the polytrauma CIG to different patients, showing

step-by-step the results of applying the CIG’s

recommendations. However, we did not have the

possibility to split the physicians attending the course

into two groups (one taking advantage of GLARE,

and the other not using it), so that we could not have

a practical analysis of the educational advantages

provided by the adoption of GLARE. However, this

has just been our first experience in adopting GLARE

(and META-GLARE) for educational purposes. Our

roadmap is much more ambitious, and it is briefly

presented in the following section.

4 ROADMAP TO A FULL

EXPLOITATION OF

META-GLARE FOR

EDUCATION

Besides the “conservative” application of GLARE for

education in the ROPHS project, in which GLARE

has been used “as it is” to show students practical

examples about how to operate on specific

(polytrauma) patients, our roadmap plans much more

ambitious applications, and consequently, a lot of

extensions. Notably, to achieve a fast prototyping of

such extensions, we will take advantage of META-

GLARE, which has been explicitly designed in order

to make extensions of CIG formalisms easy to

implement-manage.

4.1 Explanatory Annotations

For the sake of education, it is very important that

each action in the CIGs is carefully annotated with

detailed explanations and motivations. Indeed, GL

are based on best practices and medical evidence, and

such knowledge has to be explicitly stored in order to

support education. Of course, such an extension does

not require specific efforts from the computer science

point of view, but, indeed, the acquisition of such

annotation is a long and time-consuming task, to be

performed mostly by domain experts.

4.2 Propose & Check Execution

Modality

As discussed in Section 2, issue (ii), also “second-

opinion” simulation can play an important role in

medical education. However, to support “second-

opinion” simulation, GLARE execution engine must

be enriched with a new modality, in which GLARE,

instead of suggesting the most suitable choice for

diagnostic and therapeutic decisions, let the student

choose without any suggestion, and then compares

the student’s choice with the one that would have

been suggested by the CIG. Especially in case of

disagreement (but also in case the choice of the

student is the same that would have been

recommended by the GL) the motivations for the

choice recommended by the GL should be provided

to the students (e.g., by showing the logic underlying

the decision – see the discussion about therapeutic

and diagnostic decisions in Section 3.1).

4.3 Fake Decisions and Paths

By definition, GLs contain the “best practices” to

cope with specific diseases. However, during the

educational processes, diagnostic and therapeutic

problems might be made more complex for students

by adding wrong decisions leading to non-existing (in

the real GL) alternative paths into the GL (called fake

decisions and paths in the following). Fake decisions

and path can be used in the “second-opinion”

simulation, to increase the complexity of the problem,

especially in case such fake alternatives represent

cases of frequent\plausible medical errors. Of course,

the explanatory facility is fundamental in this context:

each fake path must be paired with exhaustive

annotations, detailing why such a path should not be

applied.

4.4 What-if Analysis

The “what-if” analysis in an important cognitive tool

in human reasoning in general, and in the medical

context in particular, since it allows one to analyse the

consequences of a given action or choice. GLARE

already supports a quite reach mechanism to support

what-if analysis (Terenziani et al., 2002). GLARE’s

“what-if” facility is the implementation of a form of

hypothetical reasoning. In particular, users are helped

in gathering the various types of information

necessary for discriminating among the alternative

paths of actions at any stage of the GL. Relevant

decision parameters (e.g., costs, resources, times) are

gathered from the alternative paths in an automatic

Towards an “Operational” Educational Model in Healthcare: Exploiting Computer-Interpretable Guidelines

407

way. At the end of this process, the tool displays the

values of the chosen parameters for each of the

alternative paths.

Such a mechanism can be used for education in

two different ways:

(1) In the “standard” simulation, to show to the

students the different consequences of

decision actions in the GLs.

(2) In the “second-opinion” simulation, to

support students’ choices. Before taking a

decision, students could use such a facility to

explore the consequences of choosing one

alternative or another.

4.5 Patients’ Data Generation

A central issue is the definition of patients. Indeed,

(the data of) a specific patient will be the starting

point of each educational session, in which students

will have to simulate the application of the GL “best

practices” on her\him. Since we are speaking of

education, we do not “physically” operate on real

patient. Therefore, while in a real application (of a

GL on a patient) the whole evolution of the data of

the patient cannot be known “a-priori” (but it is

discovered “step-by-step”, looking at the evolution of

the status of the patient), in the educational context

we only simulate an application, so that we can

hypothesize to know a-priori the whole evolution of

the patient. However, different forms of simulations

can be provided, depending on the model one adopts

to represent the patient data. At least two different

approaches are possible, that we call (i) deterministic

patient and (ii) probabilistic patient.

In the “deterministic patient” model, we assume

that the clinical record of the patients contains, “a-

priori” (i.e., already at the beginning of the

simulation) the whole evolution of the patient’s data.

There is only one specific evolution, corresponding to

a specific path in the CIG, the one recommended for

such a patient. A temporal (relational) database is

used to manage such temporal data (as regards

GLARE, consider, e.g., (Anselma et al., 2018; Stantic

et al., 2012)). The deterministic patient model can be

used in the “standard” simulation (since the data

evolution corresponds to the path of actions that one

wants to show to the students). It may also be used for

the “second-opinion” simulation. However, in such a

case, whenever students deviate from the

recommended path, they have to receive a warning

(and exhaustive explanations), and then be forced to

continue the simulation considering the

recommended path (since only the data

corresponding to the recommended path are available

to the system).

On the other hand, the probabilistic patient model

represents the initial status of a patient, and the

probabilistic history of her\his evolutions, depending

on the GL actions performed on her\him. Obviously,

the definition of (“probabilistic”) patients is complex,

and requires the availability of a lot of knowledge,

e.g. (Dagan et al., 2007), (Real et al., 2015). However,

it supports also flexible “second-opinion”

simulations, in which students may also follow non-

recommended paths in the CIGs.

Finally, it is important to emphasize that, while

the whole history of the evolution of the (data of the)

patients must be provided as input to the simulation,

the students will operate “step-by-step” in the

application of the CIGs to the patients, so that, at each

step, they will only see the data holding at that point

of the execution, plus past data.

4.6 Experimental Evaluation

Needless to say, the experimental evaluation is

necessarily a cue issue in the educational context. One

of the major goals we have in this context is to be able

to establish partnerships or projects with educational

institutions, in order to be able to provide extended

and accurate experimental evaluations of the impact

of adopting CIG systems in medical education.

5 CONCLUSIONS

GLs and CIGs have a quite consolidated role in the

standardization and optimization of the healthcare

services. On the other hand, their potentiality in the

education of medical students and, more generally, of

healthcare agents has not been adequately explored.

In this position paper, we claim that CIG systems

have great potentiality in educations. Since they

support the “simulation” of GLs on specific patients,

they can be used to show students how to apply

medical knowledge and best practices on specific

patients, providing them an “operational

methodology” that, otherwise, they could only learn

from the medical practice.

In this paper, we have taken GLARE as an

example of CIG system, and we have addressed the

roadmap we intend to follow in order to fully exploit

its potentialities in medical education.

HEALTHINF 2019 - 12th International Conference on Health Informatics

408

REFERENCES

Anselma, L., Piovesan, L., Stantic, B., Terenziani, P., 2018.

Representing and querying now-relative relational

medical data. Artif. Intell. Med. 86, 33–52.

https://doi.org/10.1016/j.artmed.2018.01.004

Anselma, L., Terenziani, P., Montani, S., Bottrighi, A.,

2006. Towards a comprehensive treatment of

repetitions, periodicity and temporal constraints in

clinical guidelines. Artif. Intell. Med., Temporal

Representation and Reasoning in Medicine 38, 171–

195. https://doi.org/10.1016/j.artmed.2006.03.007

Bottrighi, A., Giordano, L., Molino, G., Montani, S.,

Terenziani, P., Torchio, M., 2010. Adopting model

checking techniques for clinical guidelines verification.

Artif. Intell. Med. 48, 1–19. https://doi.org/10.1016/

j.artmed.2009.09.003

Bottrighi, A., Terenziani, P., 2016. META-GLARE: A

meta-system for defining your own computer

interpretable guideline system—Architecture and

acquisition. Artif. Intell. Med. 72, 22–41.

https://doi.org/10.1016/j.artmed.2016.07.002

Corriere, M.D., Minang, L.B., Sisson, S.D., Brancati, F.L.,

Kalyani, R.R., 2014. The use of clinical guidelines

highlights ongoing educational gaps in physicians’

knowledge and decision making related to diabetes.

BMC Med. Educ. 14, 186. https://doi.org/10.1186/

1472-6920-14-186

Dagan, I., Gabay, M., Barnea, O., 2007. Fluid resuscitation:

computer simulation and animal experiments. Conf.

Proc. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. IEEE

Eng. Med. Biol. Soc. Annu. Conf. 2007, 2992–2995.

https://doi.org/10.1109/IEMBS.2007.4352958

Field, M.J., Lohr, K.N. (Eds.), 1990. Clinical Practice

Guidelines: Directions for a New Program. National

Academies Press (US), Washington (DC).

Fridsma, D.B., 2001. Special Issue on Workflow

Management and Clinical Guidelines. J. Am. Med.

Inform. Assoc. 22, 1–80.

Gordon, C., Christensen, J. P. (Eds.), 1995. Health

telematics for clinical guidelines and protocols, Studies

in health technology and informatics. IOS Press,

Amsterdam, Netherlands.

Leonardi, G., Bottrighi, A., Galliani, G., Terenziani, P.,

Messina, A., Della Corte, F., 2012. Exceptions handling

within GLARE clinical guideline framework. AMIA

Annu. Symp. Proc. AMIA Symp. AMIA Symp. 2012,

512–521.

Montani, S., Terenziani, P., Bottrighi, A., 2005. Exploiting

Decision Theory for Supporting Therapy Selection in

Computerized Clinical Guidelines, in: Miksch, S.,

Hunter, J., Keravnou, E.T. (Eds.), Artificial Intelligence

in Medicine, 10th Conference on Artificial Intelligence

in Medicine, AIME 2005, Aberdeen, UK, July 23-27,

2005, Proceedings, Lecture Notes in Computer Science.

Springer, pp. 136–140. https://doi.org/10.1007/

11527770_19

Peleg, M., 2013. Computer-interpretable clinical

guidelines: A methodological review. J. Biomed.

Inform. 46, 744–763. https://doi.org/10.1016/j.jbi.

2013.06.009

Piovesan, L., Molino, G., Terenziani, P., 2015. Supporting

Multi-level User-driven Detection of Guideline

Interactions, in: Verdier, C., Bienkiewicz, M., Fred,

A.L.N., Gamboa, H., Elias, D. (Eds.), HEALTHINF

2015 - Proceedings of the International Conference on

Health Informatics, Lisbon, Portugal, 12-15 January,

2015. SciTePress, pp. 413–422.

Piovesan, L., Molino, G., Terenziani, P., 2014. Supporting

Physicians in the Detection of the Interactions between

Treatments of Co-Morbid Patients, in: Healthcare

Informatics and Analytics: Emerging Issues and

Trends. IGI Global, pp. 165–193.

Real, F., Riaño, D., Alonso, J. R., 2015. A Patient

Simulation Model Based on Decision Tables for

Emergency Shocks, in: Riaño, D., Lenz, R., Miksch, S.,

Peleg, M., Reichert, M., ten Teije, A. (Eds.),

Knowledge Representation for Health Care. Springer

International Publishing, Cham, pp. 21–33.

Stantic, B., Terenziani, P., Governatori, G., Bottrighi, A.,

Sattar, A., 2012. An implicit approach to deal with

periodically repeated medical data. Artif. Intell. Med.

55, 149–162.

Terenziani, P., Montani, S., Bottrighi, A., Molino, G.,

Torchio, M., 2008. Applying artificial intelligence to

clinical guidelines: the GLARE approach. Stud. Health

Technol. Inform. 139, 273–282.

Terenziani, P., Montani, S., Bottrighi, A., Torchio, M.,

Molino, G., 2002. Supporting physicians in taking

decisions in clinical guidelines: the GLARE“ what if”

facility. Proc. AMIA Symp. 772.

Terenziani, P., Montani, S., Bottrighi, A., Torchio, M.,

Molino, G., Correndo, G., 2004. A context-adaptable

approach to clinical guidelines. Stud. Health Technol.

Inform. 107, 169–173.

Towards an “Operational” Educational Model in Healthcare: Exploiting Computer-Interpretable Guidelines

409