MEDNET: A Medical Simulation Network Michael D. Myjak, M.S. The Virtual Workshop, Inc. P.O. Box 98 Titusville, FL 32781 Joseph Rosen, M.D. Dartmouth Hitchcock Medical Center One Medical Center Drive, Lebanon, NH 03756 Keywords: Advanced Distributed Simulation (ADS) DIS Telemedicine HLA Abstract War fighting technology is challenging the capability of military medical personnel to improve combat casualty care. Except for the introduction of helicopter evacuation support during the Korean War, the percentage of soldiers lost in combat has not changed significantly since World War II. While training for combat and strategic planning has improved with the advent of battlefield simulators, it is now time to apply these tools to the training for and implementation of combat casualty care. Mednet - An Overview MEDical simulation NETwork or MEDNET is a vision of a simulation system that can be used for future combat casualty care, civilian medical training, and in times of terrorist attack, to augment basic first-aid care. The MEDNET system would be a fully integrated part of the overall strategic training mission of the armed forces. The Advanced Distributed Simulation (ADS) technology upon which MEDNET is based is a blend of the Defense Modeling and Simulation Offices (DMSO) High Level Architecture (HLA), incorporates enhanced Distributed Interactive Simulation (DIS), constructive and live simulations capabilities with modern, emerging telemedical technology. For example, MEDNET could enable the simulation of civilian patients (or wounded combatants), their surroundings, or present a virtual clinician (via the Internet) to enhance the skills and capabilities of medical and military medical clinicians, or immediate first-aid care givers. The components of MedNet The concept for MEDNET is based on existing and reconfigurable simulation system technology suitable for both individual and team training. The primary components of MEDNET include the ADS system [the core of MEDNET], the virtual patient simulator, a fully immersive 3D rendering system, an injury catalog incorporating a library of branched-physiological scenarios, and an adaptive intelligent interface module called the Virtual Clinician Assistant. Using the components of MEDNET, military and civilian medical corps of the future will be able to train in a synthetic (e.g., urban street, battlefield, rural area) environment, treat virtual injuries, and seek expert guidance all from within the virtual environment of MEDNET. Whether from a portal on World Wide Web to a fully immersive virtual environment, MEDNET would be capable of offering a wide range of educational and training capabilities. Through its Web portal interface, MEDNET has the capability of providing basic first aid information to a broad constituency. In instances of terrorist attack, hundreds of thousands of people are going to want to know where to turn for aid. Local and regional hospitals and trauma centers may become quickly over loaded. Health officials may be unable to triage the vast majority of the injured public. Where will these people turn for basic information? The Internet. And where on the Internet will they be able to find up-to-the-minute information on an on-going crisis? MEDNET. In its fully immersive state, a virtual reality (VR) cave would be built around modern data-grade video graphics projectors, suspended from the ceiling and displaying on four surrounding walls. The expectation of this system is that this simulated "synthetic" environment will render a 360-degree field of view that fully immerses the participant(s). In the center of this rendered space resides a table -- a high definition volumetric display -- that will present the virtual patient image to the interactors. This display presents a stereoscopic image that appears as a scale model resting on the table. The interactor may view this image from any angle: by walking around the table or by leaning over it to gain perspective from various azimuths and altitudes. Perhaps of particular interest are the data fusion capabilities of this display, to be able to integrate graphical or multi-dimensional datasets with the virtual patient display. A true synthesis of MRI, 3D ultrasound, or other CT data can be used to simulate a particular patient or condition. When merged and morphed with data from the Visable Human Project, a true-to-life rendering is produced. MEDNET covers a wide range of the medical simulation application domain. And recent advances in the World Wide Web, hardware doubling in performance every 6-8 months, and now the Internet-2 and the Next Generation Internet projects indicate that future bandwidth will be available to support and sustain such a simulation system. In the next section, we detail several key components of the MEDNET system. The Virtual Patient Simulator At the center of the MEDNET synthetic environment will be the high-fidelity Virtual Patient Simulator or VPS. The VPS is the subsystem responsible for modeling and rendering the human patient form. The core of the human model used by MEDNET will be constructed using a blend of ADS technology developed by the U.S. Department of Defense and the Visible Human project from the National Institute of Health. Scenario-specific data for a virtual patient simulation will be initially drawn from a physiological database. Then mathematical behavioral models contained within the injury catalog will be selected for the scenario-specific event. For example, the virtual patient image generated contain generic ADS entity models. It may also be possible to extend this presentation by using a live data fed from a trauma pod or possibly from stored Magnetic Resonance Imaging (MRI) data sets. These displays will be superimposed on the digitized Visible Human and morphed as appropriate onto a standard human model. Visual or polygonalized data collected from the National Library of Medicines Visible Human project can then provide a texture mapped overlay to generate quite realistic imagery. In most technical training environments, immediate assessment and feedback on performance can greatly enhance the task acquisition process.  The VPS will be a composite simulation system (a blend of live, real-time and constructive simulation technology) utilizing and aggregation of both low and high-fidelity modeling techniques. Low fidelity modeling will be accomplished using constructive simulation techniques. When aggregated with virtual simulations, the VPS will be used to manage the majority of the patients sub-systems in a logical and coherent fashion. High fidelity modeling that requires a high-degree of interaction will be accomplished strictly using real-time distributed interactive simulation techniques. These multi-processor-based systems are capable of providing the high-degree of interaction and fidelity required to train a clinician or medical corpsman. The Haptic Interface Human-computer interaction has historically consisted of limited interaction with visual displays of iconic and character data on a two-dimensional screen. Networked Virtual Environments (Net-VEs) offer an alternative interaction paradigm in which users are no longer simply external observers of data but are active participants with their data in a 4D virtual world. Within the Net-VE, force sensation plays important rolls in recognition of 3-dimensional objects and our interaction with them. The hardware and software technology involved in the creation of interactive virtual environments such as MEDNET is still relatively new, but haptic devices are already in commercial-off-the-shelf form. A high fidelity haptic simulation of surface contact presents a demanding technical challenge in the design of force reflecting Net-VEs. In fact, the creation and quantification of the characteristics of each VE application is a research task in itself. (One of the main objectives of MEDNET is the representation of a force sensation by an interactor in the synthetic environment.) Surrounding the volumetric display within MEDNET will be a set of haptic (tactile/force-feedback) interaction tools. These tools that comprise a technology assessment will permit the interactor to reach out and touch the VPS. The haptic systems in MEDNET that permit the interactor to touch, feel, and otherwise "physically" interact with the VPS will lead to evaluation and characterization of the fidelity necessary in haptic devices and the human factors associated with tactile and force-feedback systems in medical simulation applications. High fidelity, distributed interactive simulation techniques will provide for adequate man-in-the-loop interaction and response times. Although direct feedback to the interactor will be provided for by the haptic system, interaction and rendering will be controlled by the distributed Net-VE. Entity-to-interactor interaction will encompass many various surgical tools and simulated telemedical instruments. Additional techniques can be programmed into the MEDNET system as new tools are added to support a variety of training and educational tasks. Three DIMENSIONAL Displays In many applications, the understanding and interpretation of visual images are inherent parts of the problem solving process. Examples in medical imagery range from diagnostic radiology and fluid-structure interaction to problems involving operator-assisted telerobotics. In MEDNET, the computer can be used to perform image, data, and knowledge processing in a way that is aligned with an understanding of the user. Pseudo-holographic display systems can enhance our understanding and interpretation of visual images. They also provide for more realistic imagery. For the interactors, this display system can enhance interaction with the virtual patient simulation and further the immersion effects. In addition, several different levels of medical clinicians can be trained using MEDNETs reconfigurable environment. Clinicians can perform physical assessment tasks and practice procedures. The objective of these systems is to provide high-fidelity video stimuli to the trainee with a minimal amount of distortion. This can occur in the VR Cave, or through the desktop using LCD shutter glasses. This can enhance the realism associated with the actual simulated environment. While this appears feasible and sensible on the surface, little research has been done to verify training effectiveness, cost effectiveness, human factors, or the impact of display system types. Network performance and training effectiveness are particularly troublesome and limited. However, initial system prototypes do appear quite promising, and recent assessments by the Army Research Laboratory appear to support the claim that interactors using 3D visualization appear to perform at a superior level to those using only 2D visualization. Thus we have good evidence to believe that 4D interaction may indeed be superior still. The Virtual Clinician Assistant A significant amount of research has been conducted in developing techniques for embedded assessment for intelligent tutoring systems. This area focuses on the application and extension of real-time embedded assessment technology to casualty care. Today we believe that this level of assessment could now be integrated with modern knowledge-base technology and made available to the public at-large through the Web. Consider that in most technical training environments, immediate assessment and feedback on performance can greatly enhance the acquisition process. This is especially true for procedural based applications (e.g., diagnostics, control procedures, etc.). When procedural errors are identified in real-time, it is easier for the learner to comprehend the context in which the error occurred. Often it is the situational variables that lead to a procedural error, hence corrective feedback in real-time aids in learning to avoid situationally induced errors. The Virtual Clinician is the focus of research on the development of a prototype real-time intelligent embedded assessment module that would be integrated with MEDNET. This module would capture the procedural knowledge for a selected subset of diagnostic and/or operational activities. This involves knowledge engineering of selected procedures, development of the prototype knowledge model, and integration and testing with the Virtual Clinician interface. Further, timely information could be programmed into this interface to provide Just-In-Time training to public at-large. One of the side benefits of a validated embedded assessment module is that it reduces the number of live assessment experts needed for training. The Virtual Clinician prototype will be structured so that it can be extended into a comprehensive model in the out-years of the MEDNET program. This is perhaps by far, the most visionary components of MEDNET. A key element necessary to enhance many of the tools and models being developed is the need for an intelligent diagnostic aid. The intelligent diagnostic aid would exploit neural network technology, specifically back propagation neural networks, to provide expert diagnosis based on selected physiological scenario inputs. This type of expert systems approach is vital to the development of medic and physician-centered training. A fundamental reason for the importance of this type of system to the long term goals of the MEDNET project is the same as for any expert system, the retention of expert knowledge. Often the time between armed conflicts is lengthy. As a result, each time the military enters an armed conflict, it has a staff of physicians and medics with little or no direct experience in combat casualty care. The objective of the intelligent diagnostic aid within MEDNET is to create a system that can capture combat casualty diagnostic knowledge so that it is permanently archived and accessible during future training and conflicts. Later, this knowledge can moved into the civilian sector to aid in diagnostic training and emergency room care. 7.0 Conclusion MEDNET will generate an immersive and highly adaptable virtual environment that will allow individual participants or teams to train simultaneously. The scenes presented within the MEDNET cave can change from a front line battlefield, through combat support hospital, all the way back to a remote hospital located in rural New Hampshire or urban Miami. Indeed, the entire continuum of support echelons can be modeled. Modeling treatment received prior to, during or after transportation, or post operative care can be the focus of MEDNET simulations. Situational awareness training garnered through each step in the design of the 21st century medical system will be supported.. This integrated use of MEDNET represents a training system simulation of casualty care as a comprehensive and realistic simulation exercise. The MEDNET training environment will therefore enable military and civilian medical personnel to better understand and manage the toll exacted by casualties on limited resources. 8.0 ABOUT THE AUTHORs Michael D. Myjak is Vice President of Research and Development, co-founder and Chief Technical Officer of The Virtual Workshop, Inc., where his current role is as chief architect of Javelin, a Java-based Run-Time Infrastructure for the Next Generation of Internet applications. In 1982, Mr. Myjak received two Bachelor of Science Degrees from Clemson University, one in Computer Science and the other in Engineering Technology. He obtained his Master of Science Degree in Computer Science - Systems from the University of North Texas in 1988 while employed with the Computer Science Laboratory, Corporate Research and Development labs at Texas Instruments. Prior to founding The Virtual Workshop, Mr. Myjak was a Senior Research Scientist with the Institute for Simulation and Training, at the University of Central Florida where he continues to teach "Building Virtual Worlds." Mr. Myjak has been an active participant in Modeling and Simulation standards activities for a number of years, and was recently re-elected as Chair of the Standards Activity Committee (SAC) of the Simulation Interoperability Standards Organization (SISO). He has previously Chaired of the Run Time Infrastructure and Communications Forum and the Run Time Infrastructure Interoperability Study Group under SISO, and the Internet Engineering Task Forces (IETF) Large Scale Multicast Application (LSMA) working group. Mr. Myjak is active in the Web 3D consortium's Virtual Reality Transfer Protocol Working Group, and the Internet Research Task Forces Reliable Multicast Research Group. YZ|~1>ij'&6&??7@>@BBE6]6:OJQJ5: jU5OJQJ6&'?Z}~&'0$%: ABsdHEst  +iklSTi(")"##%%''())++6/`6/7/W/00+3,3r5s566A9B9m;n;};(>)>j?k??EE$) 00P/ =!"#$%8$() 00P/ =!"#$% P8$Dd<  C A2!"o*D`!̨!"o*"?0x XUu#$?AQP 5=\b\q5P3s!%M!!q#s9r>r1yҺ.N/?RA(>π('(SBRG)EON97{yyYWW͠*)x>):bB~/‚QIkQﾫ=yT2=uQżq(.ϴ m(s²pE(G&+ɵ">Z5)vcGeˋ>Sj(䝀QQyes\lJV侾k%==]YژJ#5+_W.ήloSX^bcW9I(;&(;(;wTv~žԣGCQSm{KXUھr?Y9q)P]xD)ڲs\Գ_\wת-™iJ3}80OqCy}w[HUnY*v*~ʗc+UJ!ar'3‘ ‘̈.ŸZ7()ws7)w^ީm=^fr;Cy¡~ߠۚ.‹ɠKG?GW)y'^??**k7E+~8n+sמ_.*:w/+|tS1yPq(N.+Eӳ).CU\PS^/ʦ%K>+SU_A=5P S #XuNֆQE}? xLزPzɜL/4ܐR MIM6հwi>-EG1U>ߵ2 JYvG }GlK D"]T@76{J:d7)Ի_غ#Ӫ.~l÷JFzy,n%{7UHb٫r-?o*^7*3(e ^e#K^ԴSiS̽^ ^`KUb^uki,EIbRyҪg>^Yu>'gU*+%6?/*J7Ʋb ͗iun5kKGf:6Ԯ}%CMGjMMMuz.CG̶,4JyKEihF%m6j oQSkfj-yAbjLGiWi񂗭͛] m},m@i3zWi3Z~Mh!z z-G]y֣Kp:88f8:8 78;~t\vЧ"E~A] F Qίҟp:F^TiR}K3j465fcf,1 )\s1\d7fcݼc0yF ۨf^3'F-勌<{v/_/S3KЬ-i 4x n7r}gw3=p'o8o9l"׺a2Ea\1M/۶Y,4k:CũMlj-bYUiAcL$Ai 5awڼuqtYYÀER(EP4R'4&PgEg"tVq&9qNp:';D@ r  ʌbqJDnqfrCޙV6#\!ƵjjkZ-W/-Us˴jSWM/w㹴t^Rus!\Np~xjsnӸB vnSUo.>47գ93[zrUU%j\vJ hbLaPjE(_ ri5FZ%f }%rڞ 8_ 9r,hs"m2r$#= r"G4r!G0rzQ9CG'q9 r\r9 9. 9I9n 9:3rA 8g.rT6czRR鷻19V g96!V9v"N;_"VdF 9>r,Gf,>e|>Sˋ6||z#LRGYId%¯ 9v 9JR|ۏNF!j3k Fm #" jTjٷsRc8"Xjjs15TTy֤r,UaQBghO]y_/Kh@CT F jL@11!43c&b8R{D]c&a#&Iq.ua֙  k CHRD(EhMDRo##KBHI"r!Bh⌔s4YBGs$ F$v~ M M.'SB첢od0leY\&qL/s))Fcb )SI9D)K8T[j5LǢ3ϳi-9H5Rj% V?Xm+ Buͣ/NUr1J[Z+Qk1jZSQM}6^h}Q+VX@jHً<3k 3\?"V=PܳdZX*:**vj PEoZPj A15ʢbZZPk3j}i^%[/QK&GP֏B\ZZ#Z=»uikQ ^AeX&!2="}e|$$Z6h9v]f*&sPY\y "hs*՛fh+5ҭ䌨WhVff85V,Ւf3,HJBd$KA2 RRih6Mj6fL$d3lM%s ÔxbAf !YO$$da'(R'}d%[dKש7EԻ\*j%[(>}bkT,D(͓bU lNB|0Y})[%%[dfSm >Hf6DTt \RI<:D{iრ-1፾US B+Tv)rGG4@.\B.Esݲ]LyrU6 Jrj9sY+2 HF4fDJBDD33Ykj^+%ٳ$ԃI^egikFF9re*e::6l(sՑQF;a-6f-^b6ڳg0}UklG+#(N^7CW(3UW}jkxأtJGˉ@F^ ^{5f| .fvlwJU lVH #1kk mrxQ '5`B,2SqBLlڭĭkz}r2ɍrr= Z Mqr:G9C.?\T*,|\3"(ZQ{)WץXR"jD$)W2rpZPk2bM94iuG&&K,R+C$VjX 6׈h*zFΌ:U#!JZX+XH>Wh5lPU'*|X:-i)y)%Xu }zXKC, 2X.!V=MZ&ЀG9tv ZJUM4 XJ@~H "UsV(Zgd ʨhU :VD6$6l$IIIcС&yM*U ZE͵$2o`z 2oF>@ZEBL+V7)^hUVH?qt=ڲ@+UѭikIbVjZZEhVgh6ݹ&VLl6ݹRE1gZ-ABQ1%k]PPR狣c2Qj=JRB-![npgk)uSlyZ]Z]dzVNZ=D+YKgk,bk,fkٌ#ZJV윴V~nN |V-()Kby'JP/ TC)[D$,JZM qk2GX9d<#OZ8Zr%!WkvAHjH3k-E#E\o(*kNW4E9J-wQ u(RE*KVnaۮu:RBKlWn VGhʢZjB-j)ZSjHJUIF&ЀG9tvI䥚XP.,!"#ڹRhb% V?X+ B,|؁ b;-q"[,l?eKdKСx-1-,[llP+Q|P:jy \RK%[B*DB>bﮞ;jmT>Vj5rV jZ-P9RZZSZ Qk j@UAZPkTkjm@5naDkʴnWZsQkjM*G0&8QCuRk+SUzZ!zZ\BL)S~בr:Y"9-[تBBS!؞ CH);Yz F&K1n++C/?ح&zy ))Ǣyg|^R/? @ D F G^8:ţN m+[frjWF]jr@3sjmD>GZ'OZ*&KƇMQZCkr}\ \j3-W8e%a,;ӣ#{bi(f \(h/QT`Uktő1+L*éRU"XMijգȮBI5 ^ERh-zf;G%eK8-T؍5 G˹W4zEWz^uyWg];vTA`xz,"A!ڳ[J&2 e#F7@H^}7!5&l^k3Z^k^^ z}^^KÑBؘ5xcT Z'%z9 mTYzee^z M2^[^rmEEƈEY.Hnn bTTkP+3h>-U͸lZntP(B=j{S'VZ}GPJ#E(BA(ZQ,u8V+#H=Z#Qk$jDjGپŽ}\_ Fu פ4 ^EB*[Ƶ\6˞: QkZ!W\~\^CfuNql}!*P@-\R-ER/QTjQ5z7]3t^a[`%_ XhjM$JA:3wFf+jmӃ-^e[GuXOak#Z^Kͪ5dk<3[c D0 F@C/ف^^g`ó:J˞ΰ=Z\^KB͙UWOE-u.ف^)zz F؋5f\Z^lZ^kkF4cӾZhAk+Cu bm!Wrn^ellx SYre#qپr]r]n1r9 V-Z 1r5u"sr\-[.!u Ή(B)S4 |\q\E\Cu"jDR$JH)\^)-#p,k=>FW\k?rWz@جm;َBk#Z^kkชT\h?mC=X\bQ+F-6P'b!W΁*WuK#ٳKC/\B/EAQ&zDZezPE1_eٹJ2,g"52)ɳ)^`ǵeBP+ζ_ Pߛ0EX?-BLXPŋI,{E(z #PlM--G(H#vGڣ@ E|GH RE}ls* F Gj.$KHIRd9R2GJH%[' q/OͮuXmjjP9v7K >ȮgzsBtl@/l=De&T{V![.VUd+dhhtE/mw5f Khl A>HkHuL_U߻tR)j J06S2Zfsh (DK@~: ZKDk\heU -=22̳Rm6'K&Z2_`>1YG;TAHT ;bUʄgo!B v"t R2+T=t[5T۲Ůja:3~ӢA0ZwTjjPm6GDVVmrjMj Qm>Mm q㏓eD5xC T(G TꁂeJp_h-!3Kc#T;%T;jwJT{jlEllPP|mQmP-P6ʳbZAkjŦ'Uun|RR5󁉢Y{eT @T T \EKz$I:[& lvM%z`GX (VE_#.`LඣeFO ӲU^k;zmG]˵KMA+#W0L^u^U :\\l\hsi*bC4Kh"11t_R­:$*QYeV fZG)!˞gzDQR.B]jZZDJxCHTiTƣ9She4!&ZV!BKS[5KHIZR @9I0-YXA൤BLZMi*jHVIHHɭ7ZufjXխ3Vb4zAc9 bgi_#6= 2 2 r3^f#l' mu bC+PV ~ BhEXUX^} c+bC !VXΩSX buD>5|WXk*bF r) Za>"[#Bov@*sV"SX6)wVjul*cǀErZ-5VV ,9hZF}hu rRR]Ri==VtB:(JR8 i**GH( Rfo||RT~BB @@ J*^(AQ&ZdI(Jh%PF]qo#͸ Vc(TezRtPRy(u2Q*U l v5::DxdC&Kh^/2Ut:H }R=%mB5*-D4됋DB?uG6H0k1жk٨PB\,gBWv%Be R:dESi``K:eBPt) *C5!$)\ZPj B-G.ePjjB%!D,-R1(e6&xałS(i)CT"@q?nFV?#TBTRYE#TT9BsBBU%RlzR5@)RRqK$Di!-/#U*-U!=֖GTARp. KVB @@ D RE"^J 䀖 ֨R4(QLV d5ruB@Ul?%zJE؅k"sB,KPO;kY^bQσPP,NG X%!Dt)قdfI >B1[3o4f5,,d>!Y!Yig(aSdsZ]dحHkեi-B94Rh%(:XKʨ&HbYސ 6D[[Hv N#Y> q26t6ulYl{!ʱ!^`Cպ`Cd],,BJdřd#{hVQHV-ddKʀ ˹HʨnR9RHv!jz!_vH"p$#%!%d l1-A%-혢X5[RH6F )Y!Y$;"$ =-:J^BnV3oZBepFfd[e+[(= v`ln0gϩQfXYj{qZ %X4 r1B` F` v~F%RȒ~^K~b`ռ4ZQ,rN\+^epx`(zJ^]T HT HuZ̒UBE#RCTo0F/R#'*LR_R(Au:YJ% T/P9!sVRRWPJBR˥ *t\tLh: !vzT;d䧐Tj iLy@2<:"q+u;yOBBLQi"Y2MD$T!SgdjJ!S-0Mcx 0_&8Mp26Oױ f e<&xM6sM/2!S;dAHd Cdz"CS!:UD )\N -!z#dgdLȻТLO2C#S)hdLSi62՞ii 2-A Z-$dLJQ?*4!S:d~K嫐e3:AS*jLi[߹)uʨ,љ#юgJ"UP;P!с˨N56:Ce B9 *\6P*-/R*KpdB@TP*"P9BXY{DT})(Q|}/тU[4R" FB}HR-TR7z븊Hg~@C,܍RFHuRHuIv^UT*D+(/ri'*m:UE'otL滔LLT#S2!S#djL)BtR:-D˥b2:EaC[fn8/#cvV'Z I2C=A+'ZyU5Vu2E4J1G.G.RPBjc%nZZE"KPkZȝjeH2QkKZ9B< ʴ_ZBXzG mz֢W+Z%X5CfBxsƋY.~7S%ץK)jJCy5&WzY BZ#X3j\\U*c' Ng6m0m0m0mp3nl9lg/ D ǞD.bٷW؜b+6(v nX䘰TZuh2}(]E0X}@KART#Z4bCRkI"uG6F#8`S`G)V'*WM*l-p`l1AeM`zΩql35Mz4K ,3MYǂǂ7PbYlY|l$+XP&e򈮒|jx---XPT"F839ʲZEbRYbbŒ }F 0D` p從j%+qi/.ȉ%ػz߰ Ғ>`3G^j΁`,%W `΃{'RR'̢hlN WLCQkj† W \U"ʂB.z\U[U|+Kr5BPj\M r5Cf TLz3,c\7R;bF,tJE4 XJDAV[( eӁ\]U\r]BƲ߄fnf3fx0Օ e6_ lo W]3wr WG.ڲ-SƗy{#, ! zKzK@jUL+FG5F#8!~ƣe5f#l-\fa5{"" ',xu+*Q+S!6hC9.Vg&JgcLjyWZ3rX.u!VXw b9 y๭-V b?#VbElBXZB, ҨޒӀZ ljK(RS6bE`SsU,oN_T} }XLԭ{!-E-5grWXG+ZX mHVhXX-:X'b"VkZ%#U