(1) Steps in Development of Curriculum
-
University Core Committee for EBES 2007
-
Medical College Core Committee OCT 2007
-
EBES Curriculum Core Committee April 2008 and
-
Finalized Year wise Curriculum and Assessment Techniques in April 2009
(2) Infrastructure
-
Learning Resource Centre updated in terms of Books and literature
-
Subscribed Cochrane, TRIP databeas and EBSCO system for searching evidences
-
Wi Fi enabled Campus
-
Lecture theaters updated
(3) Faculty Training
-
Selection of Nodal and Departmental Coordinators
-
Team of faculty (6 member) deputed for 1st International Workshop on “HOW TO TEACH EBM” at PGI
-
Chandigarh
-
EBES Lecture series
-
EBES Workshop for Nodal and Departmental EBES Coordinators
-
1st International Conference on Medical Education of the Future: Strategies and Innovations for
-
Success - An Evidence Based Education System
-
1st International Workshop on “Evidence Based Education System: An Innovation in Teaching”
(4) Components and Assessment Methods (Theory and Assignments)
-
First: Ask an answerable question
-
Second: Find an Article (the evidence)
-
Third: Critically Appraise the evidence
-
(Validity, Impact, Applicability)
-
Is the study valid? (Closeness to the Truth)
-
What were the Results? (Size of Effect)
-
Does it apply to my subject, case, patient?
-
Fourth: Apply
-
Fifth: Assess
3. Results
Our findings are presented in tables 1 and 2. Batch of 1st year MBBS (2010-11) appeared in and cleared the Exam of EBES and currently it is being implemented in 2nd year MBBS. Perceived Barriers to EBE are Perceived lack of personal time, Effective literature search, undertaken at the bedside is not feasible in <10 minutes, and Computer on wheels is impossible.
Table 1. EBES Curriculum and Assessment Methods
Year (MBBS)
|
Specific Teaching Hours
|
Assessment Techniques
|
Total Marks
50
|
First
|
16
|
Theory
Assignment
|
30
20
|
Second
|
16
|
Theory
Assignment
|
30
20
|
Third
|
13
|
Theory
Assignment
|
30
20
|
Final
|
12
|
Theory
Assignment
|
30
20
|
Table 2. Faculty feedback for EBES Workshop
Session
|
Rating
|
Percentage
|
Searching Evidence
|
Very good to excellent
|
56
|
Critical Appraisal of the article
|
Very good to excellent
|
44
|
Overall teaching by resource persons
|
Very good to excellent
|
85
|
4. Discussions
We have selected EBM and EBP for three reasons: 1) There is a wide variation between what is known and what is practiced, 2) There is a wide gap between available current research knowledge and application of knowledge to patient care, and 3) These are scientifically sound, patient focused and thorough and comprehensive methods which incorporate clinical experience (10, 11, and 12). We also believe that these methods help in decreasing the level of diagnostic uncertainty enough to make optimal therapeutic decisions. EBE is a way of rationalizing behaviour and governing the practices of the teaching profession with the primary outcome of interest as improvement in academics. By EBE, learning is enhanced through variety and use of many teaching methods (13). We come to the opinion that logical and rational thinking is our birth Right. Our capacity for this is part of our original equipment, something that comes company fitted. I believe in holistic learning approach through a principle called the “50/30/20” learning philosophy. Fifty percent of the learning happens in the classrooms (one way teaching), 30% through integrated teaching with the help of problem/case based learning by incorporating evidences and 20% through self learning by using information technology (Computers). This principle helps to the students in the art of asking questions and searching for appropriate answers. However, experiential learning via role plays and quasi-reality scenario-based case studies would also be encouraged.
Although India produces 15% of world’s doctors but are they trained in a way to take-up the responsibilities of the current and future era? To me, India needs a new kind of health workforce: expert public health persons and clinicians who also understand the team philosophy, use informatics, apply evidence to the patient and thus improving the quality of care. To produce this kind of breed Medical Council of India must support Medical Colleges to take up this responsibility. Through EBES even by incorporating very few teaching modalities in our existing medical education system we will very soon achieve the target of not only HEALTH FOR ALL but also MILLENNIUM DEVELOPMENT GOALS. However, successful implementation of EBES will require radical changes to the current medical Education System to improve quality and universal access (13). According to Mahatma Gandhi,”you may never know what results come of your actions, but if you do nothing, there will be no results”. So it is better to start rightnow otherwise the cost will only rise with delay.
5. Conclusion
In this regard, so for we have sensitized the educators through inhouse, national and international workshops and also through International Conference, which is hosted by Sumandeep Vidyapeeth. We believe that it will strengthen the ties between classroom teaching and clinical setting. This system is highly effective for delivering teaching instructions as appreciated by Faculty as well as students. We are looking forward for implementation of EBES Curriculum in postgraduate courses in all constituent colleges and departments and collaboration with international experts in the field of EBE.
Acknowledgements:
The authors would like to express sincere thanks to Professor Sagun Desai, for expert comments on the manuscript.
Corresponding Author:
Dr. Suresh Kumar Rathi,
F-102, Aalekh Complex, 8, Amravati Society,
Near Yash Complex, Gotri Road, Vadodara,
PIN/ZIP code: 390021, Gujarat, India.
Tel: +91.9825449480
E-mail: rathisj@yahoo.com
References
-
Harris DL, Krause KC, Parish DC, Smith MU. Academic competencies for Medical Faculty. Fam Med 2007;39(5):343-50.
-
Freeman AC, Ricketts C. Choosing and designing knowledge assessments: Experience at a new medical school. Medical Teacher 2010;32:578-81.
-
Branch WT, Paranjape A. Feedback and reflection: Teaching methods for clinical settings. Acad.Med.2002;77(12):1185-88.
-
Singh T, Bansal P, Sharma M. A need and necessity for faculty development: the role of Medical Education Units in the Indian Context. South East Asian Journal of Medical Education 2008;2(1):2-6.
-
Garg A, Rataboli PV, Muchandi K. Students’ opinion on the prevailing teaching methods in the Pharmacology and changes recommended. Indian J Pharmacol 2004;36(3):155-58.
-
Williams G, Lau A. Reforms of undergraduate medical teaching in the United Kingdom: A triumph of Evangelism over common sense. BMJ 2004;329:92-94.
-
Sackett DL, Rosenberg WM, Gray JA, Haynes RB, Richardson WS. Evidence Based Medicine: What it is and what it in n’t. BMJ 1996; 312(7023):71-72.
-
Rosenberg W, Donald A. Evidence based medicine: an approach to clinical problem-solving. BMJ 1995; 310: 1122–26.
-
Haynes RB, Devereaux PJ, Guyatt GH. Physicians and Patients choices in evidence based practices. BMJ 2002;324(7350):1350.
-
Gruppen LD, Rana GK, Arndt TS. A controlled comparison study of the efficacy of training medical students in evidence-based medicine literature searching skills. Acad Med. 2005; 80:940-44.
-
Coomarasamy A, Khan KS. What is the evidence that postgraduate teaching in evidence-based medicine changes anything? A systematic review. BMJ 2004;329:1017.
-
Krueger PM. Teaching critical appraisal: a pilot randomized controlled outcomes trial in undergraduate osteopathic medical education. JAOA 2006; 106:656-62.
-
Dogra N, Reitmanova S, Carter-Pokras O. Twelve tips for teaching diversity and embedding it in the Medical Curriculum. Med Tech.2009;31(11): 990-93.
Original Article
Fuzzy knowledge-intensive case based classification for the detection of abnormal cardiac beats
Abdeldjalil KHELASSI 1,2, Mohamed Amin Chick 2
1. Department of Informatics, Faculty of Sciences, Tlemcen University, Tlemcen, Algeria
2. Biomedical Laboratory, Faculty of Technology, Tlemcen University, Tlemcen, Algeria
khelassi.a@gmail.com
Abstract:
This paper presents a new automated diagnostic system to classification of electrocardiogram (ECG) cardiac beats. We have developed an intensive-knowledge case based reasoning classifier which uses a distributed case base enriched by partial domain knowledge (rules). An original similarity measures is proposed by combining the sigmoid similarity function with the fuzzy sets to ameliorate the system accuracy in the detection of cardiac arrhythmias. The experiments presented in this work concern the detection of Premature Ventricular Contraction PVC, normal and abnormal cardiac beats from a pattern extracted from the Electronic medical records collected and published by Beth Israel Hospital (MIT-BIH). The achieved results demonstrate the efficiency and the performance of the developed system.
Bibliographic Information of this article:
[Abdeldjalil KHELASSI, Mohamed Amin Chick. Fuzzy knowledge-intensive case based classification for the detection of abnormal cardiac beats. Electronic Physician, 2012;4(2):565-571. Available at: http://www.ephysician.ir/2012/565-571.pdf ]. (ISSN: 2008-5842). http://www.ephysician.ir
Keywords: Classification; Intensive-knowledge case based reasoning; Fuzzy sets; similarity measures; Cardiac arrhythmia diagnosis
© 2009-2012 Electronic Physician
1. Introduction
Up till now the most important source of information used by the cardiologists for the cardiac diseases diagnosis is the Electrocardiogram (ECG). The ECG is a signal produced by an electrocardiograph, which records the electrical activity of the heart. Through its wave’s duration and axes values the cardiologists recognize the abnormality of the heart beat for detecting the cardiac arrhythmias.
The classification consists of associating an object with a predefined class. There are many methods and approaches from the artificial intelligence which prove a good performance to accomplish this task as artificial neural network ANN, fuzzy systems, and similarity based classification (SBC) and other paradigms (1-5). However, each application of these approaches has positives and weaknesses that are sometime accepted and some time not accepted according to the importance and the context of the application. The medical applications including aided diagnosis and decision support systems are a critical kind of applications where the precision and the transparency are very important because it touch the human health and life.
The CBR is an intelligent approach inspired from many discipline it draws a human reasoning model. It consists of using the prior expertise to resolve new problems. This expertise is stored as a set or collection of cases called cases base. Each case represents one problem associated with its solution. The intensive-knowledge case based reasoning is a variant of CBR in which the cases is enriched by partial domain knowledge. Also the distributed case based reasoning is a variant of CBR in which the reasoning is distributed through a set of agent and the cases through a set of case bases. These variants have been developed to ameliorate the accuracy and the performance of the systems.
In this work we have developed an original classification system (IK-CBRC) for achieving the medical applications needs and for developing a flexible and accurate model. The developed system apply the intensive-knowledge case based reasoning variant and the distributed reasoning approaches via a set of agents and a set of case bases constructed from some electronic model records EMR from the MIT-BIH database which contains a set of electrocardiogram ECG commented by BIH doctors (6). We have also combined between the fuzzy sets and the similarity measures functions to increase the accuracy of our system. To evaluate our approach we have used other EMRs from the same database. The achieved results prove the usefulness of the developed approach and the integration of the fuzzy sets in the similarity measures.
2. Materials and Methods
2.1 The Electrocardiogram ECG
The ECG is a signal produced by an electrocardiograph, which records the electrical activity of the heart over time. Through its wave’s duration and axes values they recognize the abnormal heart beat and its kind which indicates the cardiac arrhythmia (the disease). There are more than forty cardiac arrhythmias each one is characterised by some rules about the measures extracted from the ECG of the patient. The elementary unit of the Electrocardiogram ECG is the beat, as shown in Fig1, it contains six waves (P, Q, R, S, T, and U) can be also considered as four parts: the P wave, the complex QRS, the T wave and the U wave. More information about the ECG is given in (7).
Figure1.The ECG parameters
2.2 Fuzzy sets
Fuzzy sets have been introduced by Lotfi A. Zadeh (1965) as an extension of the classical notion of sets. It consists to use a degree of membership instead of a simple membership in the classic sets. In classical set theory, the membership of elements in a set is assessed in binary terms according to a -bivalent condition- an element either belongs or does not belong to the set. By contrast, fuzzy set theory permits the gradual assessment of the membership of elements in a set; this is described with the aid of a membership function valued in the real unit interval [0, 1]. Fuzzy sets generalize classical sets, since the indicator functions of classical sets are special cases of the membership functions of fuzzy sets, if the latter only take values 0 or 1 (8).
2.3The case based reasoning CBR
Case-based reasoning is a problem solving paradigm that in many respects is fundamentally different from other major AI approaches. Instead of relying solely on general knowledge of a problem domain, or making associations along generalized relationships between problem descriptors and conclusions, CBR is able to utilize the specific knowledge of previously experienced, concrete problem situations (cases) (9). The case is a contextualized piece of knowledge representing an experience that teaches a lesson fundamental to achieving the goals of the reasoning system. It is composed from two parts problem part and solution part. The main idea of case based reasoning is that two similar problems have the same solutions. Aamodt and Plaza describe the CBR life cycle as a four process summarized below (Figure 2).
F igure2. Case-Based Reasoning cycle [A.Aamodt and E.Plaza 1994] (9)
The first process consists of retrieving from the cases base the similar case or cases which can be useful to solve the current problem. In the Second process, reuse, all solutions (cases) retrieved by the retrieve process are reused to find the potential solution. The Third one called revise process; which revises and checks the solution to fit the specifics of the current problem. Finally, the retain process, which updates the memory by adding the resolved problem as a new case to the cases base. The distributed case based reasoning consists of distributing the reasoning through a set of agents and the cases through a set of case bases. There are many works in the distribution of reasoning but each one has its proper realization. Research efforts in the area of distributed CBR concentrate on the distribution of resources with the intent of improving the performance of CBR systems. Although the phrase distributed CBR can be used in a number of different contexts (10).
A knowledge-intensive case-based reasoning method assumes that cases, in some way or another, are enriched with explicit general domain knowledge. The role of the general domain knowledge is to enable a CBR system to reason with semantic and pragmatic criteria, rather than purely syntactic ones. By making the general domain knowledge explicit, the case-based system is able to interpret a current situation in a more flexible and contextual manner than if this knowledge is compiled into predefined similarity metrics or feature relevance weights. A knowledge intensive CBR method calls for powerful knowledge acquisition and modelling techniques, as well as machine learning methods that take advantage of the general knowledge represented in the system (11, 12).
2.4 Contribution
The developed classification system called IK-CBRC contains two kinds of agents Adaptation agent and Similarity agent. Each case base contains cases from the same class. Each agent uses a predefined knowledge which contains ontologies, rules and heuristics to achieve their local goals and they collaborate for achieving the global goal. Also they interact with the users with a General User Interfaces for introducing the data and the classification parameters (Data, the similarity measures function and the adaptation rules) and for applying the machine learning algorithms. The cases are enriched by a partial domain knowledge which represents the cardiologists experience mobilised by some XML rules.
The developed system combines between many intelligent approaches. First of all, the adaptation agent infer from the domain knowledge base which contains some rules defined from the doctors experiences. If the response is unknown the adaptation edit a query associated with an ontology describing its features, this task is ensured by an interactive user interface.
Figure 3. The IK-CBRC architecture
Following this, the Adaptation agent propagates the query to the similarity agents. Each similarity agent contains an ontology which describes its associated case base, via this ontology and the similarity knowledge, each similarity agent compute a rate of membership of the query in the associated class. Finally, the adaptation agent generates the solution by inferring form the adaptation knowledge applied in the results sent by the similarity agents. The rate of membership is computed by the following formula:
(1)
Where sim (Q,Ci) the global similarity between the case Ci and the query Q, N the number of cases in the case base. The global similarity is measured with deferent functions, the traditional one as the sigmoid, and the exponential is defined by Axel (5). For example the sigmoid Similarity function is defined as:
Lets Q the query and C the case, qi ci: the attribute number i respectively of the query and the case and D denotes the space of case characterization models.
sim: DxD [0,1]
(2)
Where n the number of attributes, wi the weight of the attribute Ai defined by a machine learning algorithm (gradient descent). The parameters α andare defined intuitively after some experiments, and The logarithmic distance function is defined as:
σ:DxDIR
-ln(c)-ln(q) for q,c >0
σ(qi,ci)= -ln(-c)-ln(q) for q,c <0 (3)
Undefined else
We have proposed also a new similarity metric by using the fuzzy sets and the traditional global similarity function for generating three responses 1)similar 2)unknown and 3)not similar associated with the rate of membership of each response. This original proposition increases the accuracy of the system and its transparency. The rate of membership of the similarity agents’ responses is computed with the following triangular membership functions:
(4)
(5)
(6)
Where x is the global similarity measures.
The support of the fuzzy sets (a, b) is defined intuitively by using the agents interfaces or by using a machine learning algorithm. Also the function computing the value of x is selected by the user. The rate of membership is computed by putting Us (sim(Q,Ci)) in the place of sim (Q,Ci).
3. Results
In order to estimate the usefulness of our approach, we have performed several experiments with the developed IK-CBRC software. Also for proving the impact of the proposed approach we have realized several experiments with the same data and deferent strategies. The used data are divided in two parts one for the similarity learning (450 cardiac beats) and one for the test (400 cardiac beats). In these empirical experiments, we have applied the developed classifier for the recognition of cardiac arrhythmias via the cardiac beat measures described in table1 extracted from the Electronic Medical Records EMR recorded and collected by the laboratory of BIH (Beth Israel Hospital) in Boston in the United States, which is known as the MIT-BIH data base (13).
These EMRs contains the ECG signals of some patients recorded at a frequency of 360 Hz. Two or more cardiologists have made the diagnosis for these various records and they have annotated each cardiac cycle. The extracted dataset contains some normal beats and some cardiac arrhythmias as well as premature ventricular contraction PVC. The data is obtained and calculated using an algorithm developed and implemented in the LISI laboratory at the University of Rennes 1. This algorithm is based on the technique introduced by Pan J. and Tompkins W.J (14).
Table 1: The features of the extracted pattern from the cardiac beat ECG
Attribute
|
Type
|
Description
|
Pdur
|
REAL
|
The duration of the wave P.
|
PRseg
|
REAL
|
The PR segment.
|
QRS
|
REAL
|
The QRS larger.
|
STseg
|
REAL
|
The ST segment.
|
QTInterval
|
REAL
|
The QT Interval.
|
R_priorR
|
REAL
|
Distance between the current R and the prior one.
|
R_nextR
|
REAL
|
Distance between the current R and the next one.
|
RDI
|
REAL
|
Distance between R and R’
|
AmpR_S
|
REAL
|
Distance between R and S.
|
Beat_duration
|
REAL
|
The Beat duration.
|
The following tables (2, 3 and 4) contain the recognized cardiac beats of each class in the appropriate experiments. In each experiment we have used 400 classified queries (cardiac beats) from many classes including the normal the PVC classes and others. The tested queries are taken randomly.
0>10> |