Deep Learning for Medical Decision Support Systems

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021

U. Kose et al.Deep Learning for Medical Decision Support SystemsStudies in Computational Intelligence909https://doi.org/10.1007/978-981-15-6325-6_1

1. Artificial Intelligence and Decision Support Systems

Utku Kose¹  , Omer Deperlioglu²  , Jafar Alzubi³   and Bogdan Patrut⁴  

(1)

Department of Computer Engineering, Süleyman Demirel University, Isparta, Turkey

(2)

Department of Computer Technologies, Afyon Kocatepe University, Afyonkarahisar, Turkey

(3)

Faculty of Engineering, Al-Balqa Applied University, Al-Salt, Jordan

(4)

Faculty of Computer Science, Alexandru Ioan Cuza University of Iasi, Iasi, Romania

Utku Kose (Corresponding author)

Email: utkukose@sdu.edu.tr

Omer Deperlioglu

Email: deperlioglu@aku.edu.tr

Jafar Alzubi

Email: j.zubi@bau.edu.jo

Bogdan Patrut

Email: bogdan@info.uaic.ro

The humankind has always found its way on solving problems in the real-world, by using tools and deriving solution scenarios. As the more tools designed and developed by humans, the more effective solutions and new kinds of tools for better solutions were obtained always. Eventually, the humankind started to use the concept of technology for defining all kinds of knowledge and skills employed for designing as well developing solutions for different fields [1, 2]. It is critical that the humankind started to give meaning to the life by examining it under different fields and the technologies used for making problem solutions and experiences within these fields easier—more practical. It is also important that the following features and mechanisms of the technology has been important on historical rise of the life standards [3–6]:

Technology is a dynamic concept adapting itself to the changed conditions,

Technology has not any direct—specific field scope (it may affect everything in a vertical and horizontal way),

Each technology has the ability to affect every different technology,

Technology has both theoretical aspects (knowledge) and applied aspects (skills) to make the life better,

Technology is a universal concept.

Considering the mentioned features and mechanisms, Fig. 1.1 provides a general view of the characteristics of the technology.

Fig. 1.1

Characteristics of the technology

Based on the explanations so far, it is possible to indicate that the humankind is currently surrounded by many different technologies. These technologies are generally somehow ensuring different levels of focus in the context of fields of the modern life. As associated with the current technological state of the twenty-first century, it can be expressed that both computer and communication technologies had had revolutionary changes in technological manner. As related to computers and even supportive components such as electronics, modern logic, mathematics on the background, the field of Computer Science has a long-time rise to shape the currently experienced technological state.

In the context of the Computer Science, there are different sub-fields where different theoretical and applied aspects of the Computer Science are widely discussed. As the computer technology has a triggering role in technological developments for a very long time, the role of Computer Science is important as a main research target for the scientific audience. As including the engineering aspects, Computer Science and Engineering has been an effective ground for the effective tools, and devices people using to deal with the real world, by using also power of the digital world.

Among all sub-fields of the Computer Science, artificial intelligence has a top place since it has many advantages to ensure successful applications for all fields of the modern life. As one of the most vital fields associated with the human life and the nature, the field of medical here has a strong relation with artificial intelligence. There are many different application types that can be associated with common use of artificial intelligence and medical but all these applications lead to the decision support making as humans still has the control (widely or narrowly depending on the target medical problems) at the end. As the first chapter of the book, a fresh start will be done in the context of this chapter, by providing brief explanations regarding artificial intelligence and its sub-research areas as machine/deep learning, and the role of intelligent systems in developing decision support systems, which are used within also medical.

1.1 Artificial Intelligence and Intelligent Systems

With a brief and direct definition, Artificial intelligence can be defined as the field of designing and developing systems, that can provide effective solutions for real world problems, by inspiring from human thinking—behaviors as well as actions by other living organisms and dynamics observed in the nature [7–9]. Artificial intelligence is a product by the humankind as it is very effective and efficient tools for getting automated solutions for real-world problems. As the humankind have many discoveries and inventions in the past, the artificial intelligence is the latest revolutionary invention gave a great rise to the technological developments started from middle of the twentieth century. Considering the current state of the twenty-first century, there is not any fields where artificial intelligence-based approaches, methods, and techniques are not used. In the form of just iterative code groups, it has been like a very easy task to solve difficult (and even almost impossible) problems, by employment of artificial intelligence. All these are because of some important characteristics (like the technology have as shown in Fig. 1.1) of the artificial intelligence. Figure 1.2 briefly expresses these characteristics.

Fig. 1.2

Essential characteristics of the artificial intelligence

Characteristics of the artificial intelligence has made that field effectively used tool for different fields where problems can be modeled mathematically and logically as the life itself is a typical chaos that has the meanings in terms of mathematics and the modern logic. In detail, the solution ways provided by artificial intelligence follows the chance factor as a heuristic view on getting solutions. These solutions are done in an iterative manner, by using learning steps causing an artificial intelligence-based technique-system to get enough idea about how to solve the target problem, and learn about the latest changes about the problem so that improving experience in an evolving manner. By taking the support of mathematically and logically structured algorithmic steps, all these allows multi-disciplinary applications as a result of flexibility.

Artificial intelligence is strong enough because it also has great relation with alternative fields and technologies. Solutions in the context of artificial intelligence often needs tasks to be done over target data (of the problem) so that there have been a remarkable relation with data processing approaches (image processing/signal processing), and at the end the outcomes of the artificial intelligence have become useful for introduce of different fields—technologies or sub-areas such as data mining, cybernetics, and robotics. Figure 1.3 represents a general view of that relation and the world of the artificial intelligence as generally. That view can be improved by including more relations as the artificial intelligence has a great—good relation in the context of its surroundings.

Fig. 1.3

Artificial intelligence and its relations

Currently, artificial intelligence is a typical mixture of different solution approaches, methods under these approaches, and also techniques—algorithms based on different methods. Except from detailed roots, it is possible to indicate that an artificial intelligence-based system can achieve the followings in the time of solving real-world problems [10–14]:

Pattern recognition,

Prediction/estimation,

Data discovery,

Data transformation,

Optimization,

Adaptive control,

Diagnosis.

All these problem solution ways are essential for artificial intelligence generally. But on the other hand, diversities among different techniques—algorithms and the followed methods have caused some sub-areas to be introduced under the field of artificial intelligence.

1.1.1 Areas of Artificial Intelligence

Although there are many detailed perspectives to explain sub-areas of artificial intelligence, an easier explanation can also be done accordingly. In this sense, artificial intelligence techniques are generally divided into two categories first: (1) learning techniques, (2) direct techniques. The learning techniques caused the machine learning sub-area to rise, as the most important solution approach of the artificial intelligence. In many problem solutions, the concept of machine learning is directly used for expressing the active use of artificial intelligence. Machine learning is also currently still rising with the advanced forms of techniques under deep learning [15–20]. On the other hand, direct techniques of artificial intelligence include general techniques such as fuzzy logic or natural language processing techniques and there is also intelligent optimization as the optimization-oriented solutions by the artificial intelligence [21–23]. Figure 1.4 provides a general scheme regarding to that classification/sub-areas of the artificial intelligence.

Fig. 1.4

Areas/sub-areas of artificial intelligence

1.1.2 Intelligent Systems

The concept of intelligent system may be used for defining logically a whole artificial intelligence-based system, by eliminating technical details such as which processes and algorithms are run in the system. As artificial intelligence is now at the edge of being a common thing in daily life, such general concepts are fine to be used for indicating active use of artificial intelligence-based solutions in different domains. An intelligent system can be in the form of only software or hardware supported with the software infrastructure. Intelligent system can be a problem-specific use of a certain technique or a hybrid formation with use of more than one artificial intelligence technique or combinations of both artificial intelligence techniques and alternative solutions from different fields. Turning back to the logical meaning of intelligent systems, these systems allows people to receive interactions regarding some kind of solution ways. Figure 1.5 represents a general list of these solution ways.

Fig. 1.5

Solution ways by intelligent systems

Moving from Fig. 1.5, it can be said that an intelligent system can provide some adjustive values/solutions for an instant run of another system or making that system—component better at the end. On the other hand, the intelligent system can make it work by providing a final solution to be used independently. Also, it is possible for an intelligent system to get some machine—device-oriented directives to control other electronic/electro-mechanic systems as it is seen as an interaction among machines—devices from the human view. Finally, an intelligent system can provide decision support, which means the provided solutions can be used to get a final decision or finalize a diagnosis.

1.2 Decision Support Systems

Decision making has always been a problem for people. At the time of especially critical decision making, it is important to eliminate different factors such as stress, noise, illness, or fatigue, in order to get an accurate and true decision at the end. In even decisions made in appropriate cases, different environmental factors may cause unpredicted results later. So, the subject of decision making has always been a critical topic for the research works [24–26]. At this point, the technology use has also brought many advantages to make decision making processes faster and easier, thanks to use of computer technologies and data analyzing—processing approaches. As appropriate to that, decision support systems have been developing in order to support people in different fields, for making automated decisions or at least using an assistive software system for getting advices. As general, a decision support system (DSS) can be defined as a system, which can automatically process and analyze some input data in order to reach to a decision as the output [27–29]. Because artificial intelligence is highly associated with ensuring effective and accurate DSS, such systems have been providing as in the form of intelligent systems. Figure 1.6 provides an example of a structure for a DSS [30].

Fig. 1.6

An example of decision support system [30]

Because decision making is done generally in all fields, the literature has many alternative DSS models and research works done so far. Some of remarkable fields where DSS are widely used can be listed as follows:

Finance [31–34],

Business Management [35–39],

Energy [40–44],

Education [45–47],

Environmental Engineering [48–50],

Medical [51–57].

1.2.1 Decision Support Systems for Medical and Deep Learning

Recently, there is a great need for running DSS for ensuring effective solutions for medical problems. As it was expressed before, use of artificial intelligence in the context of daily life can be meaningful for non-expert people by using logical concepts. As similar, the concept of DSS is widely used in medical applications and researched widely by the scientific audience. It is clear that the current and future state of medical with artificial intelligence includes remarkable use of DSS solutions [58–60]. As general, DSS can be used in the context of medical by focusing the related factors shown in Fig. 1.7 [58–62].

Fig. 1.7

Using factors of decision support systems in medical

As explained under the previous paragraphs, there are typical formations of DSS but in terms of medical, it is critical to meet with the listed factors. After designing the exact infrastructure of intelligent system in the context of software approach, it is easier to add frontend. Because of that, this book gave more emphasis to the solutions itself. Also, the systems explained under the next sections of the book are generally focused on disease diagnosis factor, as the processes leading to the diagnostics already employs the other factors of data-oriented works and the personal support (in terms of doctors, medical staff, and the patient) already.

Because DSS has high relations with artificial intelligence, developments in the field of artificial intelligence directly affects the way of DSS research. As a DSS requires evaluation of known samples or learning from the obtained data for having a decision finally, machine learning related methods and techniques are widely used in the context of DSS. As similar, the latest form of machine learning: deep learning has been intensively employed for designing and developing innovative DSS approaches and supporting the field of medical in this way [63–65]. The advantages of deep learning within DSS are associated with the advantages by deep learning techniques. Figure 1.8 represents these advantages generally.

Fig. 1.8

Advantages of deep learning in medical decision support systems

1.3 Summary

Artificial intelligence is a top field shaped the technological developments. As it is associated with different application types rising over different problem solution perspectives, it seems to be used more and more in the near future. Additionally, the field of artificial intelligence has already strong sub-areas as machine learning and deep learning. Thanks to the learning capabilities of machine/deep learning techniques, it is possible to solve even advanced problems and developing effective enough solutions in the context of intelligent systems, as a logical concept known by people.

This chapter made an introduction to the essentials of artificial intelligence (as within wide-narrow enough borders for this book) and then explained the concept of decision support systems with their application fields as including also medical. As the final touch, the chapter has also made a brief open for the use of deep learning in the context of medical decision support systems, and also importance of the diagnosis as followed more in this book.

After the start for essential concepts, it is better to deep inside in the next two chapters, by focusing on the widely known deep learning architectures, and then evaluating the literature including applications of deep learning architectures for medical diagnosis, as the decision support approach.

1.4 Further Learning

There are many more to discuss and express about artificial intelligence, and its roots as well as the associated topics. The interested readers in this manner are referred to [66–76] as some recent, and remarkable sources.

Currently, there are many different developments environment and even libraries for developing intelligent systems. The readers can read [77–85] in order to get some knowledge and skills about the widely used programming languages, and development perspectives.

For having some more information about decision support systems in general, the readers can read also the books [86–89].

References

1.

I. McNeil (ed.), An Encyclopedia of the History of Technology (Routledge, 2002)

2.

N. Rosenberg, R. Nathan, Exploring the Black Box: Technology, Economics, and History (Cambridge University Press, 1994)

3.

D. Edgerton, Shock of the Old: Technology and Global History Since 1900 (Profile Books, 2011)

4.

M.R. Williams, A History of Computing Technology (IEEE Computer Society Press, 1997)

5.

J.E. McClellan III, H. Dorn, Science and Technology in World History: An Introduction (JHU Press, 2015)

6.

D.R. Headrick, Technology: A World History (Oxford University Press, 2009)

7.

L. Rabelo, S. Bhide, E. Gutierrez, Artificial Intelligence: Advances in Research and Applications (Nova Science Publishers, Inc., 2018)

8.

J. Romportl, E. Zackova, J. Kelemen, Beyond Artificial Intelligence (Springer International, 2016)

9.

K. Henning, How artificial intelligence changes the world, in Developing Support Technologies (Springer, Cham, 2018), pp. 277–284

10.

D. Tveter, The Pattern Recognition Basis of Artificial Intelligence (IEEE Press, 1997)

11.

J.H. Holland, Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence (MIT Press, 1992)

12.

J. Liebowitz, Knowledge management and its link to artificial intelligence. Expert Syst. Appl. 20(1), 1–6 (2001)

13.

C. Blum, R. Groß, Swarm intelligence in optimization and robotics, in Springer Handbook of Computational Intelligence (Springer, Berlin, Heidelberg, 2015), pp. 1291–1309

14.

A. Pannu, Artificial intelligence and its application in different areas. Artif. Intell. 4(10), 79–84 (2015)

15.

Y. LeCun, Y. Bengio, G. Hinton, Deep learning. Nature 521(7553), 436–444 (2015)

16.

P. Ongsulee, Artificial intelligence, machine learning and deep learning, in 2017 15th International Conference on ICT and Knowledge Engineering (ICT&KE) (IEEE, 2017), pp. 1–6

17.

X. Du, Y. Cai, S. Wang, L. Zhang, Overview of deep learning, in 2016 31st Youth Academic Annual Conference of Chinese Association of Automation (YAC) (IEEE, 2016), pp. 159–164

18.

G. Nguyen, S. Dlugolinsky, M. Bobák, V. Tran, Á.L. García, I. Heredia, L. Hluchý, Machine learning and deep learning frameworks and libraries for large-scale data mining: a survey. Artif. Intell. Rev. 52(1), 77–124 (2019)

19.

D. Ravì, C. Wong, F. Deligianni, M. Berthelot, J. Andreu-Perez, B. Lo, G.Z. Yang, Deep learning for health informatics. IEEE J. Biomed. Health Inform. 21(1), 4–21 (2016)

20.

E. Alpaydin, Introduction to Machine Learning (MIT Press, 2020)

21.

C. Xu, Y.C. Shin, Intelligent Systems: Modeling, Optimization, and Control (CRC Press, Inc., 2008)

22.

M. Kppen, G. Schaefer, A. Abraham, Intelligent Computational Optimization in Engineering: Techniques & Applications (Springer Publishing Company, Incorporated, 2011)

23.

O. Senvar, E. Turanoglu, C. Kahraman, Usage of metaheuristics in engineering: a literature review, in Meta-Heuristics Optimization Algorithms in Engineering, Business, Economics, and Finance (IGI Global, 2013), pp. 484–528

24.

C.W. Kirkwood, Strategic Decision Making (Duxbury Press, 1997)

25.

S. Plous, The Psychology of Judgment and Decision Making (Mcgraw-Hill Book Company, 1993)

26.

E. Turban, Decision Support and Expert Systems: Management Support Systems (Prentice Hall PTR, 1993)

27.

D.J. Power, Decision Support Systems: Concepts and Resources for Managers (Greenwood Publishing Group, 2002)

28.

R.H. Bonczek, C.W. Holsapple, A.B. Whinston, Foundations of Decision Support Systems (Academic Press, 2014)

29.

D. Power, Decision support systems: from the past to the future. AMCIS 2004 Proc. 242 (2004)

30.

V. Rossi, T. Caffi, F. Salinari, Helping farmers face the increasing complexity of decision-making for crop protection. Phytopathol. Mediterr. 457–479 (2012)

31.

C. Zopounidis, M. Doumpos, Developing a multicriteria decision support system for financial classification problems: the FINCLAS system. Optim. Methods Softw. 8(3–4), 277–304 (1998)zbMATH

32.

C. Zopounidis, M. Doumpos, N.F. Matsatsinis, On the use of knowledge-based decision support systems in financial management: a survey. Decis. Support Syst. 20(3), 259–277 (1997)

33.

E. Tsang, P. Yung, J. Li, EDDIE-automation, a decision support tool for financial forecasting. Decis. Support Syst. 37(4), 559–565 (2004)

34.

H.J. von Mettenheim, M.H. Breitner, Robust decision support systems with matrix forecasts and shared layer perceptrons for finance and other applications, in ICIS (2010), p. 83

35.

A. Asemi, A. Safari, A.A. Zavareh, The role of management information system (MIS) and decision support system (DSS) for manager’s decision making process. Int. J. Bus. Manag. 6(7), 164–173 (2011)

36.

R. Sharda, S.H. Barr, J.C. MCDonnell, Decision support system effectiveness: a review and an empirical test. Manage. Sci. 34(2), 139–159 (1988)

37.

E.W. Ngai, F.K.T. Wat, Fuzzy decision support system for risk analysis in e-commerce development. Decis. Support Syst. 40(2), 235–255 (2005)

38.

V.L. Sauter, Decision Support Systems for Business Intelligence (Wiley,