IntroductionTheultimateteacheronhumanphysiologyishumanpathologyand the clinical trials aimed at i ts amelioration. However, thepathway between disease and cure most frequently arises froma reductionist approach. Perhaps no better illustration of thisprocess is the history of the diagnosis and treatment ofhemophilia A, which involved the identification ofthe defectinblood, in plasma, in cryoprecipitate [1] and ultimately by theabsenceorimpairmentofthefactor(F)VIIImolecule.Essentialto the evolution of the development of knowledge andtreatment were in vitro tests and experimental methods, theresults of which could be validated in clinical trials whilesimultaneously providing the quantitative methods essential toidentify the factor (F)VIII molecule.The activated partial thromboplastin time (APTT) test [2],still essential in hemophilia diagnosis and management, isclearly biologically artificial, employing citrate plasma and anon-physiologic activator with excess phospholipid. Nonethe-less, its utility is not in doubt. However, the relevance of somemolecular defects in hemostasis identified using the APTT areindoubtastheyarenotassociatedwithahemostaticpathology.Our laboratory, while predominately reductionist in ap-proach uses a variety of models including: numerical synthesisof the clotting reaction solved on a computer [3]; closed andopen (flow)systems utilizing the reconstituted synthetic plasmaclotting proteome [4–6]; minimally altered, corn trypsin inhib-itor-treated whole blood with and without endothelial andinflammatory cells [7,8]; and blood oozing from microvascularwoundsinvolunteers[9].Thenumericalmodelshavebeenusedin studies to discriminate between potential mechanisticpathways[10–12],epidemiologicstudiesofhemorrhagecontroland thrombosis [13] and in support of drug design andevaluation [14–16].Numerical models systems and their validationHuman biology is fundamentally opaque while numericalmodels are fundamentally transparent (Fig. 1); the gradientfrom numerical models through proteome to human physiol-ogybecomeslessandlesstransparent.Whentheresultsofthesemodels converge we believe we understand the process; whenthey diverge it becomes the basis of the discovery of newpathways. The fundamental utility of the role of numericalmodels in the long-term project of understanding coagulationhas and continues to be recognized by a number of investiga-tors at all levels of empirical model complexity. Mathematicalconstructs of the coagulation reaction network have beendeveloped using ensembles of differential equations [3,17–22]or more elaborate constructs for both closed and flow-basedmodel systems [19,23–32]. All these numerical models developwithin and from the gradient of empirical model systems usedto investigate the chemistry of coagulation. The coincidence ofnumerical and empirical proteome models suggests the modelsare correct at this level.Numerical models present a distillation of the insights fromempirical models and offer the promise of predicting outcomeswhen reaction networks are perturbed. Like the empiricalmodelsfrom whichthey derivetheircontent,numericalmodelsare both incomplete and imperfect representations of the livingsystems they attempt to describe. The gradient from totallytransparent numerical models to human physiology (Fig. 1)requires constant empirical validation with increasingly com-plex and opaque empirical studies. Any discontinuity observedin proceeding through the gradient results in the identificationof new discoveries. At each step in the progression (Fig. 1), thecosts and time for execution increases in an exponentialfashion. Numerical hypotheses tested on the computer takeseconds and cost little. Clinical trials take years and areexorbitantly expensive.Clot SpeedBiochemistry is fundamentally a reductionist science. Mylaboratorys initial foray into blood clotting dealt with theisolation and characterization of the multiple forms ofthrombin [33] and the activation of prothrombin [34]. Thesestudies subsequently proceeded through the isolation of FV[35], its activation to FVa [36], and the participation of FVa,FXa and phospholipid in the prothrombinase complex [37].The greater than 300 000-fold increase in prothrombin activa-tion rates by the FXa, FVa, phospholipid, Ca