Home About us Editorial board Search Ahead of print Current issue Archives Submit article Instructions Subscribe Contacts Login 
  • Users Online: 583
  • Home
  • Print this page
  • Email this page


 
 Table of Contents  
REVIEW ARTICLE
Year : 2016  |  Volume : 2  |  Issue : 3  |  Page : 101-112

Healthcare, uncertainty, and fuzzy logic


Command Post of Gendarme Logistics, School of Technical and Auxiliary Forces, Ankara, Turkey

Date of Web Publication24-Nov-2016

Correspondence Address:
Güney Gürsel
Command Post of Gendarme Logistics, School of Technical and Auxiliary Forces, Ankara
Turkey
Login to access the Email id

Source of Support: None, Conflict of Interest: None


DOI: 10.4103/2226-8561.194697

Rights and Permissions
  Abstract 

Fuzzy nature of decision-making process in healthcare enforces technology producers and researchers to employ creative and smooth solutions. Conversion from fuzzy concepts and ideas to crisp values causes loss of precision and weakens the output decisions. A promising bundle of techniques, soft computing, is a fast developing and popular area that helps meet this creative and smooth need in healthcare. In this study, fuzzy logic (FL) application in healthcare decision-making is examined. The number of publications is rising each year related to FL application in healthcare. FL can be used as a classifier, or in a selection process of a certain type of disease, or diseased patients, or determining the risk ratio of a disease, or in a data mining algorithm, or in constructing a decision support system. This study is a descriptive study aiming to examine and explain FL applications in healthcare.

Keywords: Decision support, fuzzy logic, healthcare, hybrid fuzzy logic applications, soft computing


How to cite this article:
Gürsel G. Healthcare, uncertainty, and fuzzy logic. Digit Med 2016;2:101-12

How to cite this URL:
Gürsel G. Healthcare, uncertainty, and fuzzy logic. Digit Med [serial online] 2016 [cited 2021 Dec 8];2:101-12. Available from: http://www.digitmedicine.com/text.asp?2016/2/3/101/194697


  Introduction Top


Softness stems from the reality that the real world is "soft," and it is not composed of white and black, one and zero, true and false. Modeling real-life situations with "hardness," translation from a continuous space into two-edged Boolean logic causes loss of valuable information and precision. [1]

Zadeh's hammer principle illustrates this situation very well, "when the only tool you have is a hammer, everything looks like a nail." [2] The saying clearly announces that you need a tool box to handle different situations. The thing to do is not only to "nail." There is a wide array of context-depending situations, so hard computing will not be able to meet all the requirements.

As an important part of the real-life applications, in medicine, it is impossible to give exact definitions or descriptions of medical concepts and relationships between concepts in most of the cases. The boundaries are not clear. Consider the statement, "If the back pain is severe and the patient is old, then apply acupuncture to certain point for a long time." To process and model this statement in a computer system, we need more than programming skills and true-false statements. All the terms we need to model that are severe, old, certain point, and long-time are vague and fuzzy. For example, size as a variable is expressed such as "very small," "small," "medium-sized," "large," "extremely large," and "monstrous." For that reason, Healthcare Information Systems applications need to employ soft computing (SC) methodologies.

Imperfect knowledge is unavoidable in medicine and the nature of medical data causes many uncertainties in medical decision-making, arising from a number of areas such as: [3]

  • Incomplete understanding of biological mechanisms
  • Imprecise test measurements
  • Uncertainty about normal ranges for test results
  • Simultaneous presence of more than one condition
  • Missing information occurring in a large percentage of cases.
Bhatia et al. [4] state one of the main problems in making medical diagnosis as classification of medical data and shows the solution in SC techniques such as Bayesian network (BN), artificial neural network (ANN), fuzzy inference system, genetic algorithms (GAs), swarm intelligence, and fuzzy cognitive maps (FCMs).

As for this, quoting Yardimci [3] will help understand the use of SC in medical domain: "Soft computing is a branch of computer science capable of analyzing complex medical data. Their potential to exploit meaningful relationship set in a data set can be used in the diagnosis, treatment and prediction of the outcome in many clinical scenarios." Rephrasing may cause the loss of the impact of the statement, so it is quoted directly.

SC is defined by Zadeh [5] as a partnership of methodologies, in which each one contributes a distinct methodology for addressing problems in the domain. He addressed the components of SC as fuzzy logic (FL), probabilistic reasoning, GAs, chaotic systems, Bayesian networks (BNs), neurocomputing, and parts of learning theory; he also stated that these partners are complementary rather than competitive.

This study is a descriptive study aiming to examine and explain FL, which is a part of the SC, and its common applications in the medical field in view of the literature. The aim of the study is to give basic information and vision about the FL usage in medical domain to handle the uncertainty and to offer decision support to the staff. In this study, because of the space limitations, some concepts, such as decision trees (DTs), ANNs, and Bayesian algorithms, are assumed to be known by the reader and not explained.

Medicine is one of the fastest growing fields when compared to other computer-aided technologies. This fast growth, together with the vague nature the medicine has, brings the need for different strategies and creative technologies such as FL or its combination with other artificial intelligence techniques.

Because of its ability to extend the classical Boolean logic of the computer applications, healthcare computer aided applications employ FL. To handle the semantics of the related domain, it compares, constraints, extends, particularizes, etc., concepts as humans do in reasoning; it connects symbols and concepts. [1]

The study is organized as follows. First, the basics of FL are given. The main part is in the section titled, "Fuzzy Logic in Medical Domain" comprising the common FL applications and techniques used in medicine. Uncertainty of medicine will also be explained in detail in this part as what is mentioned by "uncertainty." In the end, conclusions are formulated.


  Fuzzy logic basics Top


Fuzzy sets have been presented to the literature by Zadeh. [6] Seising and González [2] give the difference between hard computing and SC as follows:

  • Hard computing is rigid/precise/crisp whereas SC is flexible/approximate
  • Hard computing is b-valued whereas SC is fuzzy-valued
  • Hard computing is abstract-based whereas SC is context-based
  • Hard computing is unique whereas SC is plural
  • Hard computing uses numbers whereas SC uses words and sentences.
There are claims that FL was already present at the beginning of the century although it was first introduced in 1965 to handle vagueness, [7] by the sayings such as "I have worked out the logic of vagueness with something like completeness" in 1905 by the Philosopher Charles Sanders Peirce.

The popular terms that FL associated with are "uncertainty," "vagueness," "fuzziness," having common feature as not being clear enough to be processed by hard computing techniques.

FL deals with approximate values instead of certain values. "Certain values" are called as "crisp" in the rest of the chapter. While these certain values are cold-hot, FL values become cold-a little cold-warm-a little hot-hot.

Fuzzy systems are used in many applications in real life. Elevators, razors, subways, medical devices, bond-rating systems, risk analysis systems for bank credit applicants have shown good examples of fuzzy systems.

In the literature, there are many different definitions of FL. Drawing from literature, FL can be defined as the effort of emulating human reasoning model, which uses linguistic variables and concepts, into computer applications. In this strategy, computer is able to process linguistic variables and their degrees of memberships rather than crisp numbers and equations.

Fuzzy sets

Fuzzy sets have more than either-or approach for membership. [8] Consider the set of tall people. A man with the height of 200 cm belongs to this set and a man with the height of 110 cm does not but what about the people with the height of 150, 160, 165 cm height? Zadeh's [6] "grade of membership" solves this issue. According to this concept, a set has members belonging to it partially. [9],[10]

Considering the tall men set, everyone can make different fuzzy sets with different members. According to the man with 110 cm, 160 cm is tall whereas it is not for the man with 190 cm. This shows the fact that fuzzy sets are subjective, case dependent, and context dependent.

Especially in medical science, in most of the cases, it is impossible to give exact definitions or descriptions for medical concepts and relationships between these concepts.

Membership function

The function that assigns a number to each element x of the input space is called the membership function and represented as μ(x). The membership function maps an input value to its membership value. More clearly, membership function of x shows the degree of its membership to a fuzzy set. The values given by the membership function have to be in [0, 1] interval.

The fuzzy set Z, where Z = {(5, 0.4), (7, 0.1), (9, 1), (8, 0.2)}, is to be notated as Z = {0.4/5, 0.1/7, 1/9, 0.2/8} by fuzzy notation. Notice, membership value of zero does not appear in the set. The standard notation for the membership grade of the fuzzy set Z at 5 is

μZ (5) = 0.4.

Interpretation of the fuzziness membership determines the meaning associated with the membership function. [11]

Although the type of membership function to represent is still a research issue, the most commonly used shapes for representing the membership functions are triangular, trapezoidal, linear, and Gaussian.

The most practical and the most frequently used one is the triangular membership function. [12] According to the study of McNeill and Thro [13] and the study of Duraisamy et al., [14] triangular functions are the best one to use. Triangular membership functions are easy to use.


  Components of a fuzzy logic system Top


An FL model is a rule-based system that uses fuzzy theory to handle vagueness. [15] In [Figure 1], components of an FL system are given.
Figure 1: Fuzzy logic system components

Click here to view


Four basic components of the FL system are as follows:

  • Fuzzifier: This component is responsible from translating the crisp inputs into fuzzy values
  • Inference engine: It has the fuzzy reasoning intelligence to get the fuzzy output. Human decision-making is simulated to constitute the engine. The rules of inference are important
  • Knowledge base: It includes knowledge and decision rules, which is captured from the experts' experience of the application domain managing the relations between inputs and outputs [16]
  • Defuzzifier: It translates the fuzzy output into a crisp value. [16]
Fuzzifier

The fuzzifier is the part responsible for fuzzification. Fuzzification is the process of converting a crisp object into a fuzzy set, to a grade of membership function for linguistic variables of fuzzy sets. [7] In [Figure 2], an example of fuzzifying the cris P = 70 into the linguistic variables, low and medium, is given.
Figure 2: Membership functions of the "low" and" medium"

Click here to view


The conversion is made by intersecting the vertical line coming from 70 in the x-axis, with the horizontal line to the y-axis. The values on the y-axis are our membership values. According to the [Figure 2]

μA (70) = 0.75

μB (70) = 0.25.

Inference engine

After fuzzification, the resultant fuzzy sets (that are still input values) are processed in the inference engine according to the rules of the rule base. The inference engine is the processing unit of the FL system.

Knowledge base

This is the most important part of an FL system. The performance of an FL system depends on its knowledge base. The knowledge base of an FL system is composed of a database and a rule base.

A knowledge base can be constructed either by experts or self-learning algorithm. The first way is using experts to construct rule base. Experts of the proposed system describe the fuzzy if-then rules mentioned above. The second way is using the self-learning algorithm to construct the rule base. For these types of rule base, a part of the cases is used to train the system and construct the rule base while the other part of the cases is expected to be solved by the system. These types of self-learning systems are called neuro-fuzzy systems.

Defuzzifier

Normally, the output of the inference engine is also a fuzzy set, which is not useful in the real world. Thus, it needs to be transformed into the useful and understandable value to be used in the real world. The crisp output of the fuzzy output set should be a value, constructed by taking into account all the points in this fuzzy output interval and by employing high membership degree values more than the ones with small or no membership degree.

Advantages and disadvantages of fuzzy logic

We can list the advantages of granulation over quantization as follows: [13],[17]

  • Pretends the way in which humans interpret linguistic values
  • The transition from one linguistic value to a contiguous linguistic value is gradual rather than abrupt, resulting in continuity and robustness
  • Fewer values, rules, and decisions are required
  • More observed variables can be evaluated
  • It relates output to input, without having to understand all the variables, permitting the design of a system that may be more accurate and stable than one with a conventional control system
  • Simplicity allows the solution of previously unsolved problems
  • Rapid prototyping is possible because a system designer does not have to know everything about the system before starting work
  • They are cheaper to make than conventional systems because they are easier to design
  • They have increased robustness
  • They simplify knowledge acquisition and representation
  • A few rules encompass great complexity.
Nothing is free of course. When we make a choice, we have some drawbacks as well as winnings. Normally, FL has also some disadvantages besides its advantages. The main disadvantage is the need for extensive knowledge of the system to be modeled, which is difficult to obtain before design. The designed rule base has no guarantee to perform well. [18]

The other disadvantages can be given as follows: [19]

  • It is hard to develop a model from a fuzzy system
  • Although they are easier to design and faster to prototype than conventional control systems, fuzzy systems require more simulation and fine tuning before they are operational
  • Perhaps, the biggest drawback is the cultural bias in the United States in favor of mathematically precise or crisp systems and linear models for control systems.

  Fuzzy logic and the medical domain Top


Sadegh-Zadeh [20] states that medical knowledge is characterized by "inescapable uncertainty." Although many reasons exist for this "inescapable uncertainty," the most important reason is medicine's being inevitably vague. [20]

The complexity of medical practice results from the lack of information, imprecise information, and contradictory nature. [21] Beck and Melo [22] claim that the rise of the modern medicine in the 19 th century and rapid developments in diagnostic procedures and treatments during the 20 th century have expanded this "complexity of medical practice;" in addition, advances in digital information processing have also been said to contribute.

Physician evaluates the risk of the patient based on subjective factors and/or additional knowledge of the situation. [23] In our era of technology, almost all the physicians use an information system in diagnosing process. This information system either helps them for recording the data or supports decision-making process. In both ways, the huge volume and different types of data, it includes textual, numeric, time series, and images and makes the decision-making process more complex and uncertain. [24] Special methods are needed to handle this uncertainty. [25]

The sources of uncertainty can be classified as follows: [26]

  • Information belonging to the patient
  • Highly subjective and imprecise medical history of the patient supplied by the patient and/or his/her family
  • The physician usually obtains objective data in physical examination; however, in some cases, there is no certain boundary between normal and pathological status
  • Results of the laboratory and diagnostic tests are subject to some mistakes and improper behavior of the patient before the examination
  • Simulated, exaggerated, understated symptoms of the patient, failure to mention some of them
  • The physicians stress for the paradox of the growing number of mental disorders versus the absence of a natural classification
  • Difficulty in the classification in critical cases, particularly when a categorical system of diagnosis is considered.
Additional reasons can be listed as follows: [27]

  • Patient's symptoms are fuzzy with many related alternatives
  • Patients are not precise in expressing their situation in a practical mathematical manner, and they use ambiguous terms and language
  • Doctors have different background and experiences and may interpret in different manners
  • Pathological process presents ambiguous symptoms that are similar to those of other conditions.

  Fuzzy logic applications in the medical field Top


FL can be used alone as well as it can be used in a hybrid way with another technique(s) such as neuro-fuzzy applications, combination with NN, fuzzy-Bayesian applications, and combination with Bayesian networks.

FL has a wide area of application in medicine.

Ranking studies

Ranking studies is an important part of medicine. Determining the ranking of tests, risk factors, attributes, medical suppliers, and performance is important for decision-making process. The studies of Narasimhan and Malathi, [28] Kempowsky-Hamon et al., [29] Tadic et al., [30] Shrief et al., [31] and Dursun and Karsak [32] are some examples of ranking studies using FL methodologies in medicine.

Clustering studies

Clustering is a data mining technique. It is used for splitting the data into groups (called as clusters) to determine useful patterns for the purpose, in which objects within the same cluster have similar properties and objects of different clusters have different properties. [33] Clustering cancers, cells, genes, and images are the main areas of clustering in medicine. The studies of Thong, [34] Priya et al., [35] Sharma and Wasson, [36] and Wu et al. [37] are some examples of fuzzy clustering in medicine.

Classification studies

Like clustering, classification is also a data mining technique. In classification, we try to find group memberships for the known and predefined labels (classes). The studies of Nguyen et al., [38] Sridhar et al., [39] and Harikumar and Kumar [40] are some examples of these types of applications of FL in medicine.

Pattern recognition studies

Pattern recognition is a machine learning process to decipher the underlying patterns in the concerned subject. [41] Time series analysis tries to find patterns and rule depending on time, recognition of medical images belong to this type of studies. The studies of Banerjee et al., [42] Melin et al., [43] and Kempowsky-Hamon et al. [29] are some examples of pattern recognition applications of FL in medicine

Feature selection studies

Feature selection is the methodology that finds and eliminates the irrelevant samples in the given space of the samples to help the decider in decision-making process. This method is used especially to eliminate the healthy cells/images/tissues to spot the illness in the patients. The studies of Maji and Roy, [44] Kumar et al., [45] and Alhaddad et al. [46] are some examples.

Performance comparison studies

In these types of studies, FL methodologies are employed in comparing the performance of two or more different methods of tests/diagnosis/drugs/techniques, etc. The studies of Sharma and Kumar, [47] Wu et al., [48] and Ganasala and Kumar [49] are some examples of these types of studies.

Main application areas of FL in medicine, but not limited to, are as follows: [50]

  • Tuberculosis diagnosis
  • Clustering cancer cells
  • Cancer risk prognosis
  • Image segmentation for tumors
  • Aphasia diagnosis
  • Determination of drug dose in pharmacy
  • Heart disease diagnosis
  • Lung disease diagnosis
  • Diabetes care
  • Managing malaria disease
  • Hypothyroidism diagnosis
  • HIV infection cell determination
  • Earlier detection of arthritis and treatment
  • Anesthesia monitoring
  • Diagnosis and treatment of meningitis.
In addition to the above list, in the literature, we may see: [51]

  • Tropical diseases
  • Neurological diseases
  • Malaria diagnosis
  • Diagnosis and treatment of diabetes
  • Hepatobiliary disorder
  • Diagnosis of male impotence
  • Syndrome differentiation
  • Prostate diseases
  • Lymph diseases
  • Monitoring and control in intensive care units
  • Diagnosis of chronic obstructive pulmonary diseases
  • Diagnosis of cortical malformation.

  Medical image processing Top


Medical imaging is a strong supporting element in medical decision-making. Two-dimensional or three-dimensional medical images are generated by magnetic resonance imaging, computed tomography, digital mammography, positron emission tomography tests. These images can only be usable in medical decision support by postprocessing. There is an ongoing effort to improve image processing techniques to support medical decision-making.

FL techniques are also well used in this area. The medical domain is full of imprecise conditions and vagueness. Medical images present textures usually; there shows acquisition noise and inaccuracies in the definition of edges, so traditional techniques may not be adequate for processing. [52] Fuzzy methods appear to be more suitable for medical imaging.

Pattern recognition is the way for determining the patterns, cells, images, etc. Classification and clustering are common techniques used for pattern recognition. The major fuzzy methods in medical image pattern recognition are fuzzy clustering, fuzzy rule-based methods, fuzzy pattern-matching methods, and methods based on fuzzy relations. [53] In fuzzy clustering, fuzzy c-mean algorithm is one of the most used techniques. Although many image segmentation methods based on thresholding, clustering, region growing, etc., were introduced, they are not suitable for brain vessel region from the image captured by magnetic resonance angiography. It has been proved that the methods employing fuzzy information granulation become successful, which can be defined briefly as the concept that interprets the information as fuzzy granules. [54]

Some examples of FL-supported medical imaging applications are given below: [52]

  • Support in the diagnosis of brain tumors
  • Classification of radiographic images
  • Edges detection
  • Images thresholding
  • Classification of arrhythmias by electrocardiogram analysis.
Additional examples of FL-supported medical imaging are found in: [55]

  • Motion detection
  • Ranking segmentation paths
  • Color processing
  • Image filtering
  • Topology-preserving deformations of medical images.
Fuzzy expert systems in medicine

The computer-based tools for medical decision-making help medical staff diagnose the diseases. One of these computer-based tools that ease medical decision-making is a fuzzy expert system (FES). FESs have proven to be useful in the medical diagnosis for the quantitative analysis and qualitative evaluation of medical data, by achieving the correctness of results. [51] The development of disease-specific FES applications is the area of biggest interest to the researchers. [51]

FESs in medical domain are used for: [51]

  • Knowledge acquisition
  • Dealing with inconsistencies
  • Treatment planning
  • Advisory, monitoring, and control of systems
  • Parameters predictions
  • Artificial thinking.
Fuzzy information retrieval in medical databases

The data in medical databases are in crisp form as in all other fields of applications. To retrieve vague information from crisp databases in a fast way requires extra techniques. The literature provides three different techniques to meet this need using FL technologies:

  • Fuzzy query building
  • Employing fuzzy ontology
  • Constructing fuzzy relational databases.
Fuzzy query building in medical databases

An important application of FL in medicine relates to fuzzy queries to medical databases. To provide queries capable of meeting the requirements of the individual information needs, crisp relational query languages' level of abstraction is not adequate. [56] To build a query like "Show the low liver enzyme values for the children with celiac disease," we should have opportunity to state the linguistic concepts in the query language such as "low" or "children." The purpose of employing fuzzy queries is again dealing with impreciseness and complexity. Fuzzy queries are employed to handle imprecise and complex queries.

Till 1998, efforts have been made to extend relational query languages to fuzzy query. [56] With the help of this extension, vague selection criteria become possible and fuzzy sets can be used. Fuzzy SQL (FSQL) developed by Galindo et al. [57] is an example of an extended fuzzy query language, which is the fuzzy extended version of a standard query language. This is an ORACLE compatible fuzzy query language that allows writing flexible queries in the crisp tables using fuzzy attributes and linguistic labels. In the literature, one can find some other fuzzy extended query languages developed to handle vagueness.

Fuzzy ontology concept in the medical domain

In literature, many definitions of ontology are available. Drawing from the literature, we can define ontology as a formal vocabulary that specifies the domain knowledge as hierarchical concepts of entities, their attributes and relations, in both human and computer readable form.

Drawing from above definition, medical ontology can be defined as the conceptual representation of medical knowledge in a symbolic form. [58]

As in all crisp applications, ontologies built in a crisp concept and also suffer the same lack of flexibility and elasticity that FL has. The solution is again to employ and to deploy an FL concept into the ontology. The resulting product will be a fuzzy ontology. Precisely, in a fuzzy ontology, each entity is in relation with another entity with a degree of membership.

Fuzzy relational databases in the medical domain

As in the other techniques of FL applications, a Fuzzy Relational Database Management System (FRDBMS) is an extension of a DBMS. In this approach, the data definition in the database is made in a fuzzy linguistic way. In [Table 1], [59] we show an example of fuzzy linguistic data definition of a table. In this table, the fuzzy linguistic variable field is the linguistic counterpart of the field; LOW and HIGH represent the possible interval lowest and highest values it may get.
Table 1: Fuzzy linguistic data definition


Click here to view



  Fuzzy logic supported data mining in medicine Top


Another application area of FL in medicine is medical data mining. Data mining is used to extract information and knowledge from the huge data in healthcare and to use this information and knowledge for decreasing the costs and increasing the quality.

The two main problems in data mining are as follows: [60]

  • Heterogeneous data of various types, numerical or symbolic, precise or imprecise, ambiguous, approximate, with incomplete files
  • The complexity of the user's requests.
FL, with its capability to represent miscellaneous data, its robustness with regard to changes of the parameters of the user's environment, and its unique expressiveness, is very successful in handling with these. [60]

Almost all data mining techniques have a fuzzy version such as: [60]

  • Fuzzy comparison measures
  • Fuzzy machine learning
  • Fuzzy DTs
  • Fuzzy clustering
  • Fuzzy association rules.
Association rules are the statements of the presence/absence of items in transactions. As we stated that in the background section, the real world does not exist of clear edged objects such as white or black, true or false, they are fuzzy rather than crisp. [61] Hence, the items and transactions in the association rules should be fuzzy items and fuzzy transactions that form the fuzzy association rules. [61] In this case, a fuzzy transaction can contain more than one item corresponding to different labels of the same attribute because of the fuzzy nature. [61]

In the literature review of Kolçe and Frasheri [62] for the data mining techniques used in the diagnosis and prognosis of diseases, DTs, ANNs, and Bayesian algorithms appear to be the most well-performing algorithms for diagnosis. For prognosis, ANNs, Bayesian algorithms, DTs, and fuzzy algorithms came out to be the most well-performing algorithms. In the same study, the literature shows that in diagnosis of cancer diseases - ANNs, in heart diseases - Bayesian algorithms and in other diseases - DTs, in prognosis of cancer diseases again - ANNs, in heart diseases - ANNs and Bayesian algorithms are the most well-performing techniques.

For real-world data mining applications in medical data mining, refer to Bouchon-Meunier et al. [61]


  Fuzzy cognitive maps in medical decision-making Top


In the previous section, the fuzzy and complex nature of medical decision-making is mentioned. Systems that support medical decision-making should handle these complexity and toughness. A medical decision support system should be capable of extracting causal knowledge from the medical database, building a causal knowledge base, and making inference from it. [63]

CMs can be defined as the knowledge representation scheme that is capable of linking the symbolic approach used by medical ontologies with raw medical data that describe patient health. [58] They give cause and effect relationships easy to interpret. FCMs are the extension of CMs as fuzzy-directed graphs. The main advantage of FCMs over other methods is the dynamic properties of the FCMs, which are able to model the interactions between concepts. [58] They capture more information in the relationships between concepts, they are dynamic, combinable, and tunable, and they express hidden relationships. [64] FCM is a similar approach to human reasoning.

The toughest challenge in modeling an FCM is determining the influences of the concepts on the others and to which degree. To calculate these influences and their weights, historical data are used. We have tremendous amount of medical data in our databases. They can be used to construct the relationships and weights of the FCM models.


  Hybrid fuzzy approach in medicine Top


In the scenario that the relationships are not clear, the fuzzy if-then rules cannot be defined to support fuzzy reasoning. In these situations, techniques combine FL methods with other methodologies such as NN, GA, Bayesian reasoning, wavelet transform, and machine learning. Like combining FL with another technique, cascading FL with another technique is another method for hybrid fuzzy methods. [65] In the cascade method, two methods are applied one by another in turn.

Neuro-fuzzy combinations are mostly used in medical image processing applications. Image processing is a wide area of research. As well as neuro-fuzzy combinations, cascade methods are also used like FL and clustering and classification methods.

GA is used to capture the most optimal option by eliminating the defected possibilities. Using this optimal search of GA, GA-fuzzy hybrid approaches are used commonly in a cascade model rather than a combination. The most common usage of GA is as follows: [66]

  • Determination of membership functions with a fixed number of fuzzy rules
  • Extracting fuzzy rules with known memberships
  • Determination of both membership functions and fuzzy rules.
DTs are another form of hybridization with FL. In medical decision-making, DTs are mostly used and applied for decision support to the physician. Again, the chronic suffer of the "hard computing" exists in this method. DTs cannot handle vagueness. To eliminate this, the combination, fuzzy DTs are used. As one can guess, the weights of the DT are fuzzy sets instead of crisp numbers in the classical DT approach.

Particle swarm optimization (PSO), and advanced data mining technique, is another technique that is combined with FL in medicine. PSO is used alone in detecting cancer, especially brain tumors and breast cancer; [67] it is also used with FL in a hybrid manner. The most used medicine area is classification.

Ant colony optimization (ACO) is another advanced data mining technique also known as AntMiner and is also combined with FL in medicine like PSO. Like PSO, ACO is used for predicting crisp rules from medical databases. Contribution of FL to ACO is classically eliminating the uncertainty, vagueness, and interpretation of inexact data. Fuzzy-ACO applications are also used for classification tasks in medicine.

It is reported that neuro-fuzzy combinations are the most used hybrid technique in medicine. [3] GA-fuzzy combinations are smaller in number when compared to neuro-fuzzy combinations. [3] The distribution of the neuro-fuzzy application percentages according to the fields is given as clinical science 58%, basic science 25%, and diagnostic science 17%. [3]


  Conclusion Top


In this study, the basics of FL and its application areas in medical domain are given together with the techniques employed. The chronic disagreement between medical staff and the computer staff, in which the medical staff insists on impossibility of making a generalization about the diseases, while computer staff forces them to shape and generalize their concepts, seems coming to an end by the SC techniques. As the computer staff learn/develop elastic modeling methods and improve/employ techniques that will operate with smooth borders rather than sharp zero and one approach, they become more supportive to medical decision-making.

The literature is being increasingly interested in FL techniques in medicine. In medicine, the number of publications related to FL is increasing every year. [68] The most used methodology is FL-NN hybrid by 68%. [3] As the complexity and toughness of medical decision-making process force the industry to be more creative, it seems the researchers keep pace with this need.

The main study areas of FL appear to be ranking studies, clustering and classification studies, pattern recognition studies, feature selection studies, and performance comparison studies. The main application areas appear to be postprocessing of medical images, data mining of health data, information retrieval. It is also made clear that it can be used alone as well as used in hybrid approaches together with other SC methods. This information indicates that FL is employed in every critical decision-making process of healthcare from supply chain to diagnosis, from mining health data to retrieve information.

Fuzzy nature of the medical decision-making process makes traditional methods suffer from elasticity. We can count the gains of the FL in medicine by drawing from the literature examined as:

  • Flexibility: By taking into account all possible values including the blurred ones
  • Robustness: By making process more robust when compared to traditional techniques, with the feature of taking blurred boundaries into consideration
  • Efficiency: By using more available data.
As in all computer technologies, FL has side effects. Designing an FL system or application requires more effort and time. This makes the computation time for the desired output longer.

Medical domain is a rapid developing branch of science. Every day, new medical techniques and machines are emerging and developing. The contributing and supporting branches also have great share in this development, one of which is medical informatics. Every day, the number of researchers is increasing to support healthcare. Funds are increasing also to support these researchers and their studies. With the help of this increase, the number of studies and published papers is increasing in the literature as stated before. With the help of this increasing interest, there will be future work opportunities not only in medicine but also in contributing and supporting branches.

When the subject is medical domain, it is hard to foresee the future work in short. It will not be false if we say that the future work directions of the FL depend on the needs of medicine. As the medicine develops, new needs will arise and existing applications do not meet the new form's needs.

FRDBMS seems a newborn area and has a long way to go. Considering the fuzzy nature of medicine, we can say it will definitely develop in the future.

Image processing will definitely be another future work area. Every day, imaging techniques are developing. New expensive machines are introduced with more detailed imaging capability. Medical imaging is a huge industry with great money involved, the research funds will definitely increase. As a result, SC methodologies will get a considerable part. Fuzzy imaging techniques will develop in the future most probably.

FSQL is also a future work area candidate. As it is made compatible with the ORACLE DBMS, it is likely to see it grow in a very popular way in the future. DBMS companies will foresee a great profit in it. Together with ORACLE, the other companies will also use FSQL to compete in the healthcare institutions. Competence is one of the major contributors to science, and it will improve fuzzy database management.

As the most used technique in medicine, hybrid approaches of SC seem to have an important role in the future too. The complexity of healthcare is discussed in detail in this study, and this complexity needs cooperation of different techniques in some areas of medicine. As medicine improves, becomes more detailed, introduces more sub-detailed branches, it will not be surprising to expect more hybrid applications and techniques.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

 
  References Top

1.
Barro S, Marín R, editors. Fuzzy Logic in Medicine. Vol. 83. Berlin Heidelberg: Springer; 2002.  Back to cited text no. 1
    
2.
Seising R, González VS, editors. From hard science and computing to soft science and computing - An introductory survey. In: Soft Computing in Humanities and Social Sciences. STUDFUZZ. Vol. 273. Heidelberg: Springer; 2012. p. 3-36.  Back to cited text no. 2
    
3.
Yardimci A. Soft computing in medicine. Appl Soft Comput 2009;9:1029-43.  Back to cited text no. 3
    
4.
Bhatia A, Mago V, Singh R. Use of soft computing techniques in medical decision making: A survey. In Advances in Computing, Communications and Informatics (ICACCI, 2014 International Conference on September, 2014. IEEE. 2014. p. 1131-7.  Back to cited text no. 4
    
5.
Zadeh LA. Fuzzy logic, neural networks, and soft computing. Commun ACM 1994;37:77-84.  Back to cited text no. 5
    
6.
Zadeh LA. Fuzzy sets. Inf Control 1965;8:338-53.  Back to cited text no. 6
    
7.
Dadone P. Design Optimization of Fuzzy Logic Systems, Doctoral Dissertation, Virginia Polytechnic Institute and State University; 2001.  Back to cited text no. 7
    
8.
Jantzen J. Tutorial on fuzzy logic. Lyngby: Technical University of Denmark: Department of Automation; 1998.  Back to cited text no. 8
    
9.
Cox E, O'Hagan M, Taber R, O'Hagen M. The Fuzzy Systems Handbook with Cdrom. Orlando, FL, USA: Academic Press, Inc.; 1998.  Back to cited text no. 9
    
10.
Pedrycz W, Gomide F. An Introduction to Fuzzy Sets: Analysis and Design. Cambridge, Massachutes: MIT Press; 1998.  Back to cited text no. 10
    
11.
Bìlohlávek R, Klir GJ, editors. Concepts and Fuzzy Logic. Cambridge, Massachutes: The MIT Press; 2011.  Back to cited text no. 11
    
12.
Iveziæ D, Tanasijeviæ M, Ignjatoviæ D. Fuzzy approach to dependability performance evaluation. Qual Reliab Eng Int 2008;24:779-92.  Back to cited text no. 12
    
13.
McNeill FM, Thro E. Fuzzy Logic: A Practical Approach. Cambridge, Massachutes: Academic Press Professional; 1994.  Back to cited text no. 13
    
14.
Duraisamy V, Devarajan N, Somasundareswari D, Sivanandam SN. Comparative study of membership functions for design of fuzzy logic fault diagnosis system for single phase induction motor. Acad Open Internet J 2004;13:1-6.  Back to cited text no. 14
    
15.
Klir GJ, Folger TA. Fuzzy Sets, Uncertainty, and Information. Hemel Hampstead: Prentice-Hall; 1988.  Back to cited text no. 15
    
16.
Dweiri FT, Kablan MM. Using fuzzy decision making for the evaluation of the project management internal efficiency. Decis Support Syst 2006;42:712-26.  Back to cited text no. 16
    
17.
Bouchon-Meunier B, Zadeh LA. Fuzzy logic and its applications. France: Addison-Wesley; 1995. p. 120.  Back to cited text no. 17
    
18.
Roychowdhury A, Pratihar DK, Bose N, Sankaranarayanan KP, Sudhahar N. Diagnosis of the diseases-using a GA-fuzzy approach. Inf Sci 2004;162:105-20.  Back to cited text no. 18
    
19.
Pratihar DK, Deb K, Ghosh A. A genetic-fuzzy approach for mobile robot navigation among moving obstacles. Int J Approx Reason 1999;20:145-72.  Back to cited text no. 19
    
20.
Sadegh-Zadeh K. Fuzzy logic. In: Handbook of Analytic Philosophy of Medicine. Netherlands: Springer; 2015. p. 1055-110.  Back to cited text no. 20
    
21.
Torres A, Nieto JJ. Fuzzy logic in medicine and bioinformatics. J Biomed Biotechnol 2006;2006:91908.  Back to cited text no. 21
    
22.
Beck M, Melo S. Quality Management and Managerialism in Healthcare: A Critical Historical Survey. Basingstoke, Hampshire: Palgrave Macmillan; 2014.  Back to cited text no. 22
    
23.
Kwiatkowska M, Michalik K, Kielan K. Computational representation of medical concepts: A semiotic and fuzzy logic approach. In: Soft Computing in Humanities and Social Sciences. Berlin, Heidelberg: Springer; 2012. p. 401-20.  Back to cited text no. 23
    
24.
Hudson DL, Cohen ME. Uncertainty and complexity in personal health records. Conf Proc IEEE Eng Med Biol Soc 2010;2010:6773-6.  Back to cited text no. 24
    
25.
Shullman A. PACS/RIS/imaging. Radiology's golden age. This multi-site imaging firm improved patient care and its financial standing through RIS/PACS automation. Health Manag Technol 2009;30:12-3, 24.  Back to cited text no. 25
    
26.
Abbod MF, von Keyserlingk DG, Linkens DA, Mahfouf M. Survey of utilisation of fuzzy technology in medicine and healthcare. Fuzzy Sets Syst 2001;120:331-49.  Back to cited text no. 26
    
27.
Fujita H, Rudas IJ, Fodor J, Kurematsu M, Hakura J. Fuzzy reasoning for medical diagnosis-based aggregation on different ontologies. In: Applied Computational Intelligence and Informatics (SACI), 7 th IEEE International Symposium on 2012. IEEE. 2012. p. 137-46.  Back to cited text no. 27
    
28.
Narasimhan B, Malathi A. A Fuzzy Logic System with Attribute Ranking Technique for Risk-Level Classification of CAHD in Female Diabetic Patients. In: Intelligent Computing Applications (ICICA), International Conference on March, 2014. IEEE. 2014. p. 179-83.  Back to cited text no. 28
    
29.
Kempowsky-Hamon T, Valle C, Lacroix-Triki M, Hedjazi L, Trouilh L, Lamarre S, et al. Fuzzy logic selection as a new reliable tool to identify molecular grade signatures in breast cancer - The INNODIAG study. BMC Med Genomics 2015;8:3.  Back to cited text no. 29
    
30.
Tadic D, Stefanovic M, Aleksic A. The evaluation and ranking of medical device suppliers by using fuzzy topsis methodology. J Intell Fuzzy Syst 2014;27:2091-101.  Back to cited text no. 30
    
31.
Shrief M, Al-Atabany W, El-Wakad M. Ranking and evaluating CT departments by fuzzy logic. Int J Comput Appl 2015;122:8-15.  Back to cited text no. 31
    
32.
Dursun M, Karsak EE. Fuzzy decision approach based on QFD and FWA for selection of medical suppliers. Proc World Congr Eng 2015;2:802-7.  Back to cited text no. 32
    
33.
Bhandari P. Study of various clustering algorithms used by WEKA tool. Int J Emerging Res Manag Technol 2015;4:37-40.  Back to cited text no. 33
    
34.
Thong NT. HIFCF: An effective hybrid model between picture fuzzy clustering and intuitionistic fuzzy recommender systems for medical diagnosis. Expert Syst Appl 2015;42:3682-701.  Back to cited text no. 34
    
35.
Priya DK, Krithiga SR, Pavithra P, Kumar JR. Detection of leukemia in blood microscopic images using fuzzy logic. Int J Eng Res Sci Technol 2015;240:197-205.  Back to cited text no. 35
    
36.
Sharma S, Wasson EV. Retinal blood vessel segmentation using fuzzy logic. J Netw Commun Emerging Technol 2015;4:1-5.  Back to cited text no. 36
    
37.
Wu Y, Duan H, Du S. Multiple fuzzy c-means clustering algorithm in medical diagnosis. Technol Health Care 2015;23 Suppl 2:S519-27.  Back to cited text no. 37
    
38.
Nguyen T, Khosravi A, Creighton D, Nahavandi S. Medical data classification using interval type-2 fuzzy logic system and wavelets. Appl Soft Comput 2015;30:81.  Back to cited text no. 38
    
39.
Sridhar B, Reddy KV, Prasad AM. Mammographic image analysis based on adaptive morphological fuzzy logic CAD system. Int J Biomed Eng Technol 2015;17:341-55.  Back to cited text no. 39
    
40.
Harikumar R, Kumar PS. Fuzzy techniques and aggregation operators in classification of epilepsy risk levels for diabetic patients using EEG signals and cerebral blood flow. J Biomater Tissue Eng 2015;5:316-22.  Back to cited text no. 40
    
41.
Karmakar N. Generation of Authentic Signature-sets Utilizing Case-Based Reasoning Techniques (Doctoral dissertation); 2014.  Back to cited text no. 41
    
42.
Banerjee T, Keller JM, Popescu M, Skubic M. Recognizing complex instrumental activities of daily living using scene information and fuzzy logic. Computer Vision and Image Understanding; 2015.  Back to cited text no. 42
    
43.
Melin P, Castillo O, Kacprzyk J. Design of Intelligent Systems Based on Fuzzy Logic, Neural Networks and Nature-inspired Optimization. Vol. 601. Switzerland: Springer; 2015.  Back to cited text no. 43
    
44.
Maji P, Roy S. Rough-fuzzy clustering and unsupervised feature selection for wavelet based MR image segmentation. PLoS One 2015;10:e0123677.  Back to cited text no. 44
    
45.
Kumar S, Kashyap M, Saluja A, Bhattacharya M. Segmentation of Cotton Bolls by Efficient Feature Selection Using Conventional Fuzzy C-Means Algorithm with Perception of Color. In: Proceedings of the Second International Conference on Computer and Communication Technologies. India: Springer; 2015. p. 731-41.  Back to cited text no. 45
    
46.
Alhaddad MJ, Mohammed A, Kamel M, Hagras H. A genetic interval type-2 fuzzy logic-based approach for generating interpretable linguistic models for the brain P300 phenomena recorded via brain-computer interfaces. Soft Comput 2015;19:1019-35.  Back to cited text no. 46
    
47.
Sharma MK, Kumar A. Performance Comparison of Brain Emotional Learning-Based Intelligent Controller (BELBIC) and PI Controller for Continually Stirred Tank Heater (CSTH). In: Computational Advancement in Communication Circuits and Systems. India: Springer; 2015. p. 293-301.  Back to cited text no. 47
    
48.
Wu CH, Chiu RK, Wang SA. A cloud-based fuzzy expert system for the risk assessment of chronic kidney diseases. Int J Bus Syst Res 2015;9:315-33.  Back to cited text no. 48
    
49.
Ganasala P, Kumar V. Multi-scale decomposition transform based approach for CT and MR image fusion. J Med Imaging Health Inform 2015;5:715-23.  Back to cited text no. 49
    
50.
Prasath V, Lakshmi N, Nathiya M, Bharathan N, Neetha NP. A Survey on the applications of fuzzy logic in medical diagnosis. Int J Sci Eng Res 2013;4:1199-203.  Back to cited text no. 50
    
51.
Sikchi SS, Sikchi S. Fuzzy expert systems (FES) for medical diagnosis. Int J Comput Appl 2013;63:7-16.  Back to cited text no. 51
    
52.
Comas DS, Meschino GJ, Pastore JI, Ballarin VL. A survey of medical images and signal processing problems solved successfully by the application of type-2 fuzzy logic. J Phys 2011;332:012030.  Back to cited text no. 52
    
53.
Ara Begum S, Mema Devi O. Fuzzy algorithms for pattern recognition in medical diagnosis. Assam Univ J Sci Technol 2011;7:1-12.  Back to cited text no. 53
    
54.
Kobashi S, Hata Y, Hall LO. Fuzzy information granulation of medical images. Blood vessel extraction from 3-D MRA images. In: Fuzzy Logic in Medicine, Studies in Fuzziness and Soft Computing. Vol. 83. Heidelberg: Physica-Verlag; 2002. p. 18-35.  Back to cited text no. 54
    
55.
Kerre EE, Nachtegael M, editors. Fuzzy Techniques in Image Processing. Vol. 52. Heidelberg, Germany: Springer; 2000.  Back to cited text no. 55
    
56.
Tuben U, Becks A, Fathi M, Tresp C. Fuzzy Queries in Relational Medical Databases. FUSION'98 International Conference; 1998.  Back to cited text no. 56
    
57.
Galindo J, Medina JM, Pons O, Cubero JC. A server for fuzzy SQL queries. In: Flexible Query Answering Systems. Berlin, Heidelberg: Springer; 1998. p. 164-74.  Back to cited text no. 57
    
58.
Froelich W, Wakulicz-Deja A. Mining temporal medical data using adaptive fuzzy cognitive maps. In: Human System Interactions, 2009. HSI'09. 2 nd Conference on May, 2009. IEEE; 2009. p. 16-23.  Back to cited text no. 58
    
59.
Bodenreider O. Medical Ontology Research. Technical Report, Lister Hill National Center for Biomedical Communications; 2001. Available from: http://www.mor.nlm.nih.gov/pubs/pdf/2001-MOR-BoSC.pdf. [Last retrieved on 2016 Aug 05].  Back to cited text no. 59
    
60.
Mala I, Akhtar P, Zia SS, Mirza SH. Application of Fuzzy Relational Databases in Medical Informatics. In: Multitopic Conference (INMIC), 2011 IEEE 14 th International. IEEE; December 2011. p. 41-4.  Back to cited text no. 60
    
61.
Bouchon-Meunier B, Detyniecki M, Lesot MJ, Marsala C, Rifqi M. Real-world fuzzy logic applications in data mining and information retrieval. In: Fuzzy Logic. Berlin, Heidelberg: Springer; 2007. p. 219-47.  Back to cited text no. 61
    
62.
Delgado M, Marín N, Sánchez D, Vila MA. Fuzzy association rules: General model and applications. IEEE Trans Fuzzy Syst 2003;11:214-25.  Back to cited text no. 62
    
63.
Kolçe E, Frasheri N. A Literature Review of Data Mining Techniques Used in Healthcare Databases. Paper Presented at the ICT Innovations 2012, Ohrid, Macadonia; September, 2012.  Back to cited text no. 63
    
64.
Papageorgiou EI. Learning algorithms for fuzzy cognitive maps - A review study. IEEE Trans Syst Man Cybern C Appl Rev 2012;42:150-63.  Back to cited text no. 64
    
65.
Mahfouf M, Abbod MF, Linkens DA. A survey of fuzzy logic monitoring and control utilisation in medicine. Artif Intell Med 2001;21:27-42.  Back to cited text no. 65
    
66.
Wang L, Yen J. Extracting fuzzy rules for system modeling using a hybrid of genetic algorithms and Kalman filter. Fuzzy Sets Syst 1999;101:353-62.  Back to cited text no. 66
    
67.
Hedeshi NG, Abadeh MS. An expert system working upon an ensemble PSO-based approach for diagnosis of coronary artery disease. In: Proceedings of the 18 th Iranian Conference of Biomedical Engineering (ICBME'11). 2011. p. 249-54.  Back to cited text no. 67
    
68.
Dev U, Sultana A, Saha D, Mitra NK. Application of fuzzy logic in medical data interpretation. Bangladesh J Sci Ind Res 2015;49:137-46.  Back to cited text no. 68
    


    Figures

  [Figure 1], [Figure 2]
 
 
    Tables

  [Table 1]


This article has been cited by
1 Data Allocation Mechanism for Internet-of-Things Systems With Blockchain
Wendy Yanez,Redowan Mahmud,Rami Bahsoon,Yuqun Zhang,Rajkumar Buyya
IEEE Internet of Things Journal. 2020; 7(4): 3509
[Pubmed] | [DOI]
2 SAR image change detection method based on intuitionistic fuzzy C-means clustering algorithm
Deshuai Yin,Rui Hou,Junchao Du,Liang Chang,Hongxuan Yue,Liusheng Wang,Jiayue Liu,Weiping Zhang
Journal of Intelligent & Fuzzy Systems. 2020; 38(4): 3595
[Pubmed] | [DOI]
3 Personalized Pancreatic Cancer Management
Alison Bradley,Robert van der Meer,Colin McKay
Pancreas. 2019; 48(5): 598
[Pubmed] | [DOI]
4 Application of correlation analysis in cytology: Opportunities to study specific activity of follicular thyrocytes
O. I. Ryabukha,I. M. Dronyuk
Regulatory Mechanisms in Biosystems. 2019; 10(3): 345
[Pubmed] | [DOI]
5 Computational intelligence techniques for medical diagnosis and prognosis: Problems and current developments
Afzal Hussain Shahid,M.P. Singh
Biocybernetics and Biomedical Engineering. 2019;
[Pubmed] | [DOI]
6 Diseases diagnosis using fuzzy logic methods: A systematic and meta-analysis review
Hossein Ahmadi,Marsa Gholamzadeh,Leila Shahmoradi,Mehrbakhsh Nilashi,Pooria Rashvand
Computer Methods and Programs in Biomedicine. 2018; 161: 145
[Pubmed] | [DOI]



 

Top
 
 
  Search
 
Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
 Related articles
Access Statistics
Email Alert *
Add to My List *
* Registration required (free)

 
  In this article
Abstract
Introduction
Fuzzy logic basics
Components of a ...
Fuzzy logic and ...
Fuzzy logic appl...
Medical image pr...
Fuzzy logic supp...
Hybrid fuzzy app...
Conclusion
Fuzzy cognitive ...
References
Article Figures
Article Tables

 Article Access Statistics
    Viewed4146    
    Printed56    
    Emailed0    
    PDF Downloaded50    
    Comments [Add]    
    Cited by others 6    

Recommend this journal


[TAG2]
[TAG3]
[TAG4]