A fuzzy logic system for home elderly people monitoring (EMUTEM)

Proceedings of the 10th WSEAS International Conference on FUZZY SYSTEMS A Fuzzy Logic System For Home Elderly People Monitoring (EMUTEM) Jerome BOUDY2 , Bernadette DORIZZI2 Hamid MEDJAHED1,2 , Dan ISTRATE1 2 EPH 1 LRIT Telecom Sud Paris ESIGETEL 9 rue Charles Fourier, 91011 Evry 1 rue du Port de Valvins, 77215 France Avon-Fontainebleau cedex, France {hamid.medjahed,dan.istrate}@esigetel.fr {jerome.boudy,bernadette.dorizzi}@int-edu.eu Abstract: The purpose of this paper is to present a view of a telemonitoring system based on fuzzy logic for distress situations detection of elderly people living alone in the home environment. This system includes three different sensors: physiological sensors (cardiac frequency, activity or agitation, posture and fall detection sensor), microphones and infrared sensors. Taking into account the difficulty of the statistical modeling of abnormal situation and considering that the Fuzzy Logic has been actually used with success in the development of classifier and analysis of control systems, its use in the decision fusion module of our home elderly people monitoring system is presented and evaluated in this paper. Key–Words: Fuzzy logic, Fuzzy set, Data fusion, Health Science,Fuzzy Control. 1 Introduction we propose a new multimodal system called EMUTEM [4][5] (Environnement Multimodal pour la Télévigilance Medicale) for in home health care monitoring. EMUTEM gathers three subsystems which have been technically validated from end to end, through their hardware and software. The first one is Anason subsystem [5] with its set of microphones that allow sound remote monitoring of the acoustical environment of the elderly. The second subsystem is RFpat [5], a wearable device fixed on the elderly person, that can measure physiological data like heart rate, activity, posture and an eventual fall done by the person. The last subsystem is a set of infrared sensors called Gardien [5], that detect the presence of the person in a given part and also the person’s standing posture. World life expectancy has increased more than two fold over the past two hundred years [1]. The trend toward an increasingly aged population creates a growing demand for home care services, and more and more eledrly prefer to live alone at home. Based on this fact a significant burden for our society are expected, because to leave older peoples at home without putting them in danger, requires intelligent reliable remote monitoring systems that are able to offer them more independence and dignity than would be possible. In fact, a statistical study indicates that the elderly are at greater risk of trips and falls, in 1999, there were 204, 424 fall related A&E admissions, 39% of which were accounted for by patients 75 years or older, 25% were between the ages of 70-74 and 18% were 65 to 69 years of age [2]. 75% of those aged 75 years and over suffer from chronic disease, characterized as a long term health complication that current medical practice cannot overcome, with around 25% of these suffering from a combination of three or more such conditions[3]. For this reason various projects and consortia have been funded and created under European Community coordination, including CompanionAble1 project2 the one in which The three subsystem data streams have been separately processed with suitable algorithms to abnormal situations detection. In order to maximize correct classification performance between normal and distress situations, data fusion over the different sensors types is studied. The area of data fusion has generated great interest among researchers in many science and engineering disciplines. We have identified two major classes of fusion techniques: (1) Those that are based on probabilistic models (such as Bayesian reasoning [6] and the geometric decision reasoning), but their performance is limited when the data are too complex, therefore the model is uncontrollable. (2) Those based on connectionist models (such as neu- 1 The research leading to these results has received funding from the European Community’s Seventh Framework Programme (FP7/2007-2011) under grant agreement n.216487. 2 www.companionable.net ISSN: 1790-5109 69 ISBN: 978-960-474-066-6 Proceedings of the 10th WSEAS International Conference on FUZZY SYSTEMS tributes are difficult to measure accurately or difficult to quantify numerically. At that time, it is natural to use fuzzy sets to describe the value of these parameters. The attributes are linguistic variables, whose values are built with adjectives and adverbs of language: large, small, medium etc...and as an illustrating example, we found the recognition system proposed by Mandal et al.[17]. Some methods are based on a discretization of the attributes space which is a language. Thus a numerical scale of length will be replaced by a set of fuzzy labels, for example (very small, small, medium, large, extra large), and any measure of length, even numerical is converted on this scale. The underlying idea is to work with the maximum granularity, i.e. the minimum accuracy. ral networks MLP [7] and SVM [8]) which are very powerful because they can model the strong nonlinearity of data but with complex architecture, thus lack of intelligibility. Based on those facts and considering the complexity of the data to process (audio, physiologic and multisensory measurements) plus the difficulty of the statistical modeling of abnormal situation, fuzzy logic has been found useful to be the decision module of our multimodal monitoring system. Fuzzy logic can gather performance and intelligibility and it deals with imprecision and uncertainty. It has a history of application for clinical problems including use in automated diagnosis [9], control systems [10], image processing [11] and pattern recognition [12]. Some experts find it easier to map their knowledge onto fuzzy relationships than onto probabilistic relationships between crisply defined variables. The organization of this paper is as follows. A brief state of art about fuzzy logic in data fusion is presented in section II. Then, the configuration and the methodology of the developed detection of distress situations using fuzzy logic are explained in detail in section III. Experimental results are shown in section IV. Finally, a conclusion of this work is given in section V. 2 ◦ Class representation: Class don’t create a clear partition of the data space, but a fuzzy partition where recovery is allowed will be better adapted. A significant number of fuzzy patterns recognition methods, are just an extension of traditional methods based on the idea of fuzzy partition for example the fuzzy cmeans algorithm [18]. Historically, the idea of fuzzy partition was first proposed by Ruspini [19] in 1969. Rather than creating new methods of fusion and classification based on entirely different approaches, fuzzy logic fit naturally in the expression of the problem of classification, and tend to make a generalization of the classification methods that already exist. Taking onto account the four steps of a recognition system proposed by Bezdek et Pal [20], fuzzy logic is very useful for these steps. State of the art The use of fuzzy logic in EMUTEM fusion is motivated by two main raisons from a global point of view: ◦ Firstly the characteristic of data to merge which are measurements obtained from different sensors, thus they could be imprecise and imperfect. These data will be classified into two class normal situation and distress one, that are fuzzy because there is no clear limit between them. ◦ Data description: Fuzzy logic is used to descript syntactic data [21], numerical and Contextual data, conceptual or rules based data [22] which is the most significant contribution for the data description. ◦ Secondly, the history of fuzzy logic proves that it is used in many steps which are necessary for a data classification application. 2.1 ◦ Analysis of discriminate parameters: In image processing, there are many techniques based on fuzzy logic for segmentation, detection, contrast enhancement [23] and extraction [24]. Fuzzy logic and classification systems In 1965, Zadeh [13] introduced the concept of Fuzzy set theory. Historically, this was closely related to the concept of fuzzy measure, proposed just after by Sugeno [14]. Similar attempts at proposing fuzzy concept were also made at the same time by Shafer (evidence theory [15]) and Shackle (surprise theory [16]). Since that time, fuzzy logic has been more studied, and several applications were developed, essentially in Japan. The use of fuzzy sets can be done mainly at two levels: ◦ Attributes representation: It may happen that data are uncomplete or noisy, unreliable, or some atISSN: 1790-5109 ◦ Clustering algorithms: The aim of these algorithms is to label a set of data into C groups, so that obtained groups contain the most possible similar individuals. Fuzzy c-mean algorithm and fuzzy ISODATA [25] algorithm are the famous in this category. ◦ Design of the discriminator: The discriminator is designed to produce a fuzzy partition or a clear one, describing the data. This partition corresponds to classes. Indeed the fuzzy ISODATA algorithm will be adapted for this step. 70 ISBN: 978-960-474-066-6 Proceedings of the 10th WSEAS International Conference on FUZZY SYSTEMS 2.2 How to do fuzzy logic and y2 is C2 and...and yp is Cp . Where Ai and Ci are fuzzy sets that define the partition space. The conclusion of a Mamdani rule is a fuzzy set. It uses the algebraic product and the maximum as Tnorm and S-norm respectively, but there are many variations by using other operators. Fuzzy logic reflects human reasoning based on inaccurate or incomplete data. It uses the concept of partial membership, each element belongs partially or gradually to fuzzy sets that have been already defined. And in contrast to classical logic where the membership function u(x) of an element x belonging to a set A could take only two values: uA (x) = 1 if x ∈A or uA (x) = 0 if x 6∈A, fuzzy logic introduces the concept of membership degree of an element x to a set A and uA (x) ∈[0, 1], here we speak about truth value. ◦ Takagi/Sugeno rules [26]: If x1 is A1 and x2 is A2 and...and xp is Ap Then y = b0 + b1 x1 + b2 x2 + ... + bp xp . In the Sugeno model the conclusion is numerical. The rules aggregation is in fact the weighted sum of rules outputs. ◦ Defuzzification: The last step of a fuzzy logic system consists in turning the fuzzy variables generated by the fuzzy logic rules into real value again which can then be used to perform some action. There are different defuzzification methods,in the EMUTEM decision module we could use Centroid of area (COA), Bisector of area (BOA), Mean of Maximum (MOM), Smallest of Maximum (SOM) and Largest of Maximum (LOM). Equations 1, 2, 3 and 4 illustrate them. Figure 1: Fuzzy inference system steps. Figure 1 shows the main fuzzy inference system steps that are used in EMUTEM decision module which are: ◦ Fuzzification: First step in fuzzy logic is to convert the measured data into a set of fuzzy variables. It is done by giving value (these will be our variables) to each of a set membership functions. Membership functions take different shapes as it is on the figure 2. The choice of function shape is determinate iteratively, according to type of data and taking into account the experimental results. ZCOA ZBOA = xM ; Pn uA (xi )xi = Pi=1 n M X (1) i=1 uA (xi ) uA (xi ) = i=1 n X uA (xj ) (2) j=M +1 PN ZM OM = ∗ i=1 xi N ZSOM = min(x∗i ) and ZLOM = max(x∗i ) (3) (4) where x∗i (i = 1, 2, ..., N ) reach the maximal values of uA (x). 3 The main advantages of using fuzzy logic system are the simplicity of the approach and the capacity of dealing with the complex data acquired from the three subsystems: Anason, RFpat and Gardien. Fuzzy set theory offers a convenient way to do all possible combinations with these data. Fuzzy set theory is used in this system to determine the most likely distress situations that might occur for elderly persons in their home. The data fusion is carried out at three different levels: for sound at decision level, for infrared system at input data level and for wearable physiological sensor at representation level. Figure 2: Membership functions. ◦ Fuzzy rules and inference system: The fuzzy inference system uses fuzzy equivalents of logical AND, OR and NOT operations to build up fuzzy logic rules. There are several types of fuzzy rules, we mention only the two mains used in our system: ◦ Mamdani rules [26] which is of the form: If x1 is A1 and x2 is A2 and...and xp is Ap Then y1 is C1 ISSN: 1790-5109 Application 71 ISBN: 978-960-474-066-6 Proceedings of the 10th WSEAS International Conference on FUZZY SYSTEMS 3.1 Methodology and configuration The first step for implementing this approach is the fuzzification of system outputs and inputs obtained from each subsystem. From Anason subsystem three inputs are built. The first one is the sound environment classification, where all sound class and distress expressions detected are labeled on a numerical scale according to their alarm level. Four membership functions are set up in this numerical scale according to the following fuzzy levels: no signal, normal, possible alarm and alarm as it is shown in figure 3. Two other inputs are associated to each SNR calculated on each microphone (two microphones are used in the current application), and these inputs are split into three fuzzy levels: low, medium and high. Figure 4: Fuzzy sets defined for the input variables produced by RFpat. Figure 3: Fuzzy sets defined for the input variables produced by Anason. Figure 5: Fuzzy sets defined for the input variables produced by Gardien. As seen in figure 4, RFpat produce five inputs: Heart rate for which three fuzzy levels are specified normal, possible alarm and alarm; Activity which has four fuzzy sets: immobile, rest, normal and agitation; Posture is represented by two membership functions standing up / setting down and lying; Fall and call have also two fuzzy levels: Fall/Call and No Fall/Call. The defined area of each membership function associated to heart rate or activity is adapted to each monitored elderly person. For each infrared sensor a counter of motion detection with three fuzzy levels (low, medium, high) is associated, and a global one for all infrared sensors. As we use six vertical infrared sensors and two horizontal infrared sensors, Gardien subsystem delivers nine other inputs to the fuzzy inference module. The last input which is time is shown in figure 6, it has two membership functions day and night which are also adapted to patient habits. ISSN: 1790-5109 Figure 6: Fuzzy sets defined for the time input. The EMUTEM fuzzy inference component has two outputs, an alarm output with two fuzzy levels (normal and distress situation) and a localization output where the classical areas of a house are the fuzzy levels. Figure 7 shows the membership functions of these outputs, Gaussian functions is chosen for the alarm outputs, and trapezoid functions for the localization output. We have chosen the Gaussian function 72 ISBN: 978-960-474-066-6 Proceedings of the 10th WSEAS International Conference on FUZZY SYSTEMS 3.2 for the alarm output in order to obtain also a confidence level of each alarm rule necessary for a better alarm decision. Membership functions of localization are ordered according to the repartition of the classical areas in the house. Fuzzy rules and defuzzification EMUTEM fuzzy inference engine is formulated by two groups of fuzzy IF-THEN rules. One group controls the output variable localization according to values of the input variables infrared sensors Ci and SNR of each microphone. The other group controls the output variable alarm according to all inputs. An example of fuzzy rule for alarm detection is: ◦ If (Anason classification is no signal) and (Heart rate is possible alarm) and (Activity is immobile) and (Cc is low) and (C8 is low) and (Cg is low) Then( Alarm is alarm) A confidence factor is accorded for each rule and to aggregate these rules we have the choice between Mamdani or Sugeno approaches available under our fuzzy logic component. After rules aggregation the defuzzification is performed by the smallest value of maximum method for the alarm output and the centroid of area for the output localization. 3.3 Implementation Figure 7: Fuzzy sets defined for the output variables. Figure 8 shows the in-home sensors disposal and the repartition of the home into areas in order to evaluate and to supervise the EMUTEM experiments. Microphones have been calibrated using a standard level sinusoidal signal with a frequency of 1 KHz. Infrared sensors are also calibrated and for each one a specified monitoring zone is delimited. Figure 9: Software architecture of EMUTEM system. Figure 9 provides a synoptic block-diagram scheme of the software architecture of the EMUTEM system, it is implemented under Labwindows CVI and C++ software. It is developed in a form of design component. We can distinguish three main components, the acquisition module, the synchronization module and the fuzzy inference component. It can run on off-line by reading data from a data base or online by processing in real time data acquired via the acquisition module. To avoid the loss of data, a real time module with two multithreading tasks is integrated in the synchronization component. The EMUTEM system is now synchronized on Gardien subsystem be- Figure 8: In-home sensors disposal. ISSN: 1790-5109 73 ISBN: 978-960-474-066-6 Proceedings of the 10th WSEAS International Conference on FUZZY SYSTEMS cause of his smallest sampling rate (2 Hz) and periodicity. We have developed a data fusion based Fuzzy tools which allow the easy configuration of input intervals of Fuzzyfication, the writing of fuzzy rules and the configuration of the defuzzyfication method. The general interface of the system allows to build up membership functions of inputs and outputs and displaying them. We could also write rules on text file by using a specific language, that we have developed, understandable by our system. The software implementation is validated with many experimental tests. The results and the rules which produced them will be displayed on this main panel. 4 used in this fuzzy inference system. The used strategy consisted in realizing several tests with different combination rules, and based on obtained results one rules are added to the selected set of rules in order to get the missed detection. With this strategy good results are reached for the two outputs (about 95% of good detection for Alarm generation and 97% of good localisation ); the system contains 10 rules for alarm output and 16 rules for localization output. We have started the tests of the 20 scenarios using basic fuzzy rules which suppose that if a modality or several indicate an alarm the system output is also an alarm (in this case, there are not contradictory inputs). As the progress of the tests, we’ve added other rules or deleted in order to solve contradictory situations, taking into account the confidence level of inputs (sound analysis send a confidence measure). These first results encourage us to further tests in real time. In the framework of CompanionAble project will we benefit of the knowledge of medical partners about possible distress situations. Experimental results In order to demonstrate the effectiveness of this software, firstly we started by using simulated data in order to validate each rule. This simulation gaves very promising results for the alarm generation and localization without any false alarm. Figure 10 shows results for a stream of data. After the validation of each 5 Conclusion The main purpose of this paper is to present a new fusion architecture based on fuzzy logic for in-home elderly remote monitoring. The EMUTEM system which encloses this architecture is implemented and validated by simulation and experimentation. Experimental results were accurate and robust. The fuzzy logic decision module reinforces the secure detection of older person’s distress events and his localization. This approach allows easiest combination between data and adding other sensors. This constitutes a great asset of EMUTEM system to offer the possibility in a next future to implement very intelligent remote monitoring system in care receiver houses. Acknowledgements: The authors gratefully acknowledge the contribution of European Community’s Seventh Framework Programme (FP7/2007-2011), CompanionAble Project (grant agreement n. 216487). References: Figure 10: Results for a stream of data. [1] J. Riley, Rising Life Expectancy, A Global History, Cambridge University Press, New York, USA, 1995. rule, the EMUTEM system was tested with 20 scenarios, 10 scenarios with distress situations and 10 normal scenarios. These reference scenarios are based on real situations and they aim to reflect the elderly person’s everyday life. This fist study devoted to the evaluation of the system by taking into account rules ISSN: 1790-5109 [2] P. Scuffham, S. Chaplin and R. Legood, Incidence and costs of unintentional falls in older people in the United Kingdom, In Journal of Epidemiology and Community Health, Volume 57, pp. 740–744, 2003. 74 ISBN: 978-960-474-066-6 Proceedings of the 10th WSEAS International Conference on FUZZY SYSTEMS [17] D.P. Mandal, C.A. Murthy, and S.K. Pal, Formulation of a multivalued recognition system, IEEE Tr. on Systems, Man and Cybernetics, 1992 ,22:607–620. [18] W. Pedrycz, Fuzzy sets in pattern recognition: methodology and methods. Pattern Recognition, 23(1/2):121-146, 1990. [19] E. H. Ruspini, A new approach to clustering, Inform, Control, 15(1):22-32, 1969. [20] J.C. Bezdek and S.K. Pal (eds),Fuzzy Models for Pattern Recognition, IEEE Press, 1992. [21] M. Mizumoto, J. Toyoda and K. Tanaka, General formulation of formal grammars, Info Sci., 4:87-100, 1972. [22] Y.H. Pao, Adaptive Pattern Recognition and Neural Networks, Addison-Wesley, 1989. [23] J.M. Keller, R. Krishnapuram, Fuzzy set methods in computer vision, In R.R. Yager and L.A. Zadeh, editors, An Introduction to Fuzzy logic Applications in Intelligent Systems, Kluwer Academic, 1992, pp. 121–145. [24] S.K. Pal and B. Chakraborty, Fuzzy set theoretic measure for automatic feature evaluation, IEEE Trans. on Systems, Man and Cybern., 16:754-760, 1986. [25] J.C. Dunn, A fuzzy relative of the isodata process and its use in detecting compact wellseperated clusters.J. Cybernetics, pp. 32–57, 1973. [26] J.-S. R. Jang, C.-T Sun, E. Mizutani, NeuroFuzzy and Soft Computing : A Computational Approach to Learning and Machine Intelligence. Prentice Hall Upper Saddle River, NJ 1997. [3] UK Department of Health. The expert patient: A new approach to chronic disease management of the 21st century, September 2001 [4] H. Medjahed, D. Istrate, J. Boudy, F. Steenkeste, J.L. Baldinger and B. Dorizzi,A Multimodal Database for a Home Remote Medical Care Application, International conference in medical biometrics ICMB, 4-5 January , 2008,Hong Kong, pp. 99–106. [5] H. Medjahed, D. Istrate, J. Boudy, F. Steenkeste, J.L. Baldinger, I. Belfeki, V. Martins and B. Dorizzi, A Multimodal Platform for Database Recording and Elderly People Monitoring, Funchal-Madeira, Portugal, 28-31 January 2008, pp. 385–392. [6] R. Cowell, A. Dawid, S. Lauritzen, and D. Spiegelhalter, Probabilistic Networks and Expert Systems, 1999, ISBN : 0-387-98767-3. [7] G. Dreyfus, J.M Martinez, M. Samuelides, M. Gordon, F. Badran, S. Thiria and L. Hrault, Rseaux de neurones. Mthodologie et applications,Eyrolles, 2002. [8] C.J.C. Burges, A tutorial on SVM for Pattern Recognition. Data Mining and Knowledge Discovery, volume 2, 1998, pp. 121–167. [9] K.P Adlassnig, Fuzzy set theory in medical diagnosis, IEEE Tr. On Syst.,Man, and Cybernetics, March/April 1986,pp. 260–265. [10] D. Mason, D. Linkens and N. Edwards, Selflearning fuzzy logic control in medicine, Proc. AIME’97, (E. Keravnou et al., eds.), Lecture Notes in Artificial Intelligence 1211, SpringerVerlag, Berlin 1997, pp. 300–303. [11] A. Lalande, et al., ). Automatic detection of cardiac contours on MR images using fuzzy logic and dynamic programming, Proc. AMIA Ann. Fall Symp. 1997, pp. 474–478. [12] G. Zahlmann, M. Scherf and A. Wegner, A neurofuzzy classifier for a knowledgebased glaucoma monitor, Proc. AIME’97, (E. Keravnou et al., eds.), Lecture Notes in Artificial Intelligence 1211, Springer-Verlag, Berlin 1997, pp. 273–284. [13] L.A. Zadeh, Fuzzy sets as a basis for theory of possibility, Fuzzy Set Systems 1978, pp. 3–28. [14] M. Sugeno, Theory of fuzzy integrals and its applications, Doct. Thesis, Tokyo Institute of technology 1974. [15] G. Shafer, A Mathematical Theory of Evidence, Princeton Univ. Press 1979. [16] G.L. Shackle, Decision, Order and Time in Human Affairs, Cambridge Univ. Press ISSN: 1790-5109 View publication stats 75 ISBN: 978-960-474-066-6