Nof S.Y. Springer Handbook of Automation

Подождите немного. Документ загружается.

Automation and Control in Biomedical Systems References 1375

76.4.3 Model and Network Structure

Identiﬁcation

How can one develop a mechanistically accurate model

when limited measurements are available? This is si-

multaneously one of the greatest advantages, and most

severe limitations, of systems biology. Through mathe-

matical models, a large combination of model structures

representing observed behaviors can be quickly con-

structed and tested with a particular data set. However,

the challenge is in identifying which model is cor-

rect mechanistically, to the degree that the data set can

provide such information [76.99]. These model-based

methods allow extrapolation toestimate unknown quan-

tities, but can also be used to close the loop on (systems)

biology [76.100]. Close collaboration between experi-

mental biologists, clinicians, and modelers can provide

the feedback required to explore the predictions of

systems biology models from the intracellular to the or-

ganism scale. Experimental resolution deﬁnes the scale

of thestudy, and simultaneouslyprovides context for in-

terpretation and bounds limiting extrapolation. Models

relate the perturbation to the observed responses, and

an event may be represented by a wide range of models

or even model classes. It is at this point that unify-

ing frameworks, as discussed in [76.100,101], provide

engineering judgement in the construction of complex

system models.

76.5 Summary

From an automation perspective, the context in which

models will be used plays an important role [76.38].

In order to employ a model-based algorithm in a clin-

ical setting, the model must be robust enough to capture

clinically relevant behavior and the algorithm needs to

reliably return a clinically deployable treatment that

beneﬁts the patient. Based on the increasing complex-

ity of the underlying biosystems models, in terms of

both structure and computational complexity, the al-

gorithmic needs for treatment design approaches are

signiﬁcant. While these model-based treatment algo-

rithms will not change the role of the clinician in quite

the same way that computer-controlled systems have

affected the chemical plant operator, a mathematically

rigorous approach to biosystems analysis and disease

treatment has the potential to beneﬁt patient treatment –

the endpoint objective of automating biomedical treat-

ment decisions.

References

76.1 S.A.P. Haddad, R.P.M. Houben, W.A. Serdijn: The

evolution of pacemakers, Eng. Med. Biol. Mag. 25,

38–48 (2006)

76.2 L.A. Geddes: Historical highlights in cardiac pacing,

Eng. Med. Biol. Mag. 2, 12–18 (1990)

76.3 R.H. Taylor: Medical Robotics and Computer-

Integrated Surgery, 2nd edn. (Wiley, New York

1999) pp. 1213–1227, Chap. 65

76.4 T. Teorell: Kinetics of distribution of substances ad-

ministered to the body I, Arch. Int. Pharmacodyn.

Ther. 57, 202–225 (1937)

76.5 T. Teorell: Kinetics of distribution of substances ad-

ministered to the body II, Arch. Int. Pharmacodyn.

Ther. 57, 226–240 (1937)

76.6 B. Ramakrishnan, U.R. Müller: High sensitivity ex-

pression proﬁling. In: Microarray Technology and

Its Applications, ed. by U.R. Müller, D.V. Nicolau

(Springer, Berlin, Heidelberg 2005) pp. 229–250

76.7 U.S. Government: Code of Federal Regulations –

Title 21 – Food and Drugs (2008)

76.8 L. Xu, J.L. Eiseman, M.J. Egorin, D.Z. D’Argenio:

Physiologically-based pharmacokinetics and mo-

lecular pharmacodynamics of 17-(allylamino)-17-

demethoxygeldanamycin and its active metabolite

in tumor-bearing mice, J. Pharmacokinet. Phar-

macodyn. 30, 185–219 (2003)

76.9 F.J. Doyle III, D. Seborg, R.S. Parker, B.W. Bequette,

A. Jeffrey, X. Xia, I. Craig, T.J. McAvoy: A tutorial

on biomedical process control, J. Proc. Control 17,

571–594 (2007)

76.10 J.A. Florian Jr.: Modeling and Dose Schedule Design

for Cycle-Speciﬁc Chemotherapeutics. Ph.D. Thesis

(Univ. Pittsburgh, Pittsburgh 2008)

76.11 S. Audoly, L. D’Angio, M.P. Saccomani, C. Co-

belli: Global identiﬁability of linear compartmental

models – a computer algebra algorithm, IEEE Trans.

Biomed. Eng. 45, 36–47 (1998)

76.12 S. Audoly, G. Belu, L. D’Angio, M.P. Saccomani,

C. Cobelli: Global identiﬁability of nonlinear mod-

els of biological systems, IEEE Trans. Biomed. Eng.

48, 55–65 (2001)

76.13 J.W.T. Yates: Structural identiﬁability of phys-

iologically based pharmacokinetic models, J.

Pharmacokinet. Pharmacodyn. 33, 421–439 (2006)

Part H 76

1376 Part H Automation in Medical and Healthcare Systems

76.14 N.A.W. van Riel, E.D. Sontag: Parameter estima-

tion in models combining signal transduction and

metabolic pathways: the dependent input ap-

proach, IEEE Proc. Syst. Biol. 153, 263–274 (2006)

76.15 G. Bellu, M.P. Saccomani, S. Audoly, L. D’Angio:

DAISY: a new software tool to test global identi-

ﬁability of biological and physiological systems,

Comput. Methods Programs Biomed. 88,52–61

(2007)

76.16 E. Carson, C. Cobelli: Modelling Methodology for

Physiology and Medicine (Academic, San Diego

2001)

76.17 MATLAB: The MathWorks (Natick 2008)

76.18 D.Z. D’Argenio, A. Schumitzky: ADAPT II Users

Guide: Pharmacokinetic and Pharmacodynamic

Systems Analysis Software. Biomedical Simulations

Resource (Univ. Southern California, Los Angeles

1997)

76.19 P.H.R. Barrett, B.M. Bell, C. Cobelli, H. Golde,

A. Schumitzky, P. Vicini, D.M. Foster: SAAM II: sim-

ulation, analysis, and modeling software for tracer

and pharmacokinetic studies, Metab. Clin. Exp. 47,

484–492 (1998)

76.20 E.R. Carson, C. Cobelli, L. Finkelstein: The Math-

ematical Modelling of Metabolic and Endocrine

Systems (Wiley, New York 1983)

76.21 L.B. Sheiner, T.M. Ludden: Population phar-

macokinetics/dynamics, Annu. Rev. Pharmacol.

Toxicol. 32, 185–209 (1992)

76.22 S.L. Beal, L.B. Sheiner: NONMEM User’s Guide. NON-

MEM Project Group (Univ. California, San Francisco

1992)

76.23 Resource Facility for Population Kinetics,

http://depts.washington.edu/rfpk (accessed March

17, 2008)

76.24 J.A. Florian Jr., M.J. Egorin, W.C. Zamboni, J.L. Eise-

man, T.F. Lagattuta, C.P. Belani, G.S. Chatta, H.I.

Scher, D.B. Solit, R.S. Parker: A physiologically-

based pharmacokinetic (PBPK) and pharmacody-

namic model of Docetaxel (Doc) and neutropenia in

humans, Am. Soc. Clinical Oncology Annu. Meeting

(Chicago, 2007)

76.25 R.B. Conolly, M.E. Andersen: Biologically based

pharmacodynamic models: tools for toxicological

research and risk assessment, Annu. Rev. Pharma-

col. 31, 503–523 (1991)

76.26 H. Kitano: Systems biology: a brief overview, Sci-

ence 295, 1662–1664 (2002)

76.27 H. Kitano: Computational systems biology, Nature

420, 206–210 (2002)

76.28 R.E. Kronauer, G. Gunzelmann, H.P.A. Van Dongen,

F.J. Doyle III, E.B. Kelrman: Uncovering physiologic

mechanisms of circadian rhythms and sleep/wake

regulation through mathematical modeling, J.

Biol. Rhythms 22, 233–245 (2007)

76.29 H.Kitano,K.Oda,T.Kimura,Y.Matsuoka,M.Csete,

J. Doyle, M. Muramatsu: Metabolic syndrome and

robustness tradeoffs, Diabetes 53, S6–S15 (2004)

76.30 M. Hucka, A. Finney, B.J. Bornstein, S.M. Keat-

ing, B.E. Shapiro, J. Matthews, B.L. Kovitz,

M.J. Schilstra, A. Funahashi, J.C. Doyle, H. Ki-

tano: Evolving a lingua franca and associated

software infrastructure for computational systems

biology: the systems biology markup language

(SBML) project, IEE Proc. Syst. Biol. 1,41–53(2004)

76.31 L. Strömbäck, V. Jakoniene, H. Tan, P. Lambrix:

Representing, storing and accessing molecular in-

teraction data: a review of models and tools, Brief.

Bioinform. 7, 331–338 (2006)

76.32 D.P. Bertsekas: Dynamic Programming and Opti-

mal Control, Vols 1, 2 (Athena Scientiﬁc, Belmont

1995)

76.33 R. Martin, K.L. Teo: Optimal Control of Drug Admin-

istration in Cancer Chemotherapy (World Scientiﬁc,

River Edge 1994)

76.34 F. Allgöwer, T.A. Badgwell, J.S. Qin, J.B. Rawl-

ings, S.J. Wright: Nonlinear Predictive Control and

Moving Horizon Estimation – An Introductory

Overview. Advances in Control – Highlights of ECC

’99 (Springer, London 1999) pp. 391–449

76.35 K.R. Muske, J.B. Rawlings: Model predictive control

with linear models, AIChE J. 39(2), 262–287 (1993)

76.36 E. Zaﬁriou, H.-W. Chiou: Output constraint soft-

ening for SISO model predictive control, Proc. Am.

Control Conf. (San Francisco 1993) pp. 372–376

76.37 J.H. Lee, M.S. Gelormino, M. Morari: Model predic-

tive control of multi-rate sampled-data systems:

a state-space approach, Int. J. Control 55(1), 153–

191 (1992)

76.38 R.S. Parker, F.J. Doyle III: Control–relevant mod-

eling in drug delivery, Adv. Drug Deliv. Rev. 48,

211–228 (2001)

76.39 J.A. Florian Jr., R.S. Parker: Feedback Control in

Drug Delivery. In: Nanotechnology in Therapeu-

tics: Current Technology and Applications,ed.by

N.A. Peppas, J.Z. Hilt, J.B. Thomas (Horizon Bio-

science, Norwich 2007) pp. 25–64, Chap. 2

76.40 V.W. Bolie: Coefﬁcients of normal blood glucose

regulation,J.Appl.Physiol.16, 783–788 (1961)

76.41 R.N. Bergman, L.S. Phillips, C. Cobelli: Physiologic

evaluation of factors controlling glucose tolerance

in man, J. Clin. Invest. 68, 1456–1467 (1981)

76.42 R.S. Parker, F.J. Doyle III, N.A. Peppas: A model-

based algorithm for blood glucose control in type

I diabetic patients, IEEE Trans. Biomed. Eng. 46(2),

148–157 (1999)

76.43 R. Hovorka, V. Canonico, L.J. Chassin, U. Haueter,

M. Massi-Benedetti, M.O. Federici, T.R. Pieber,

H.C. Schaller, L. Schaupp, T. Vering, M.E. Wilinska:

Nonlinear model predictive control of glucose con-

centration in subjects with type 1 diabetes, Physiol.

Meas. 25, 905–920 (2004)

76.44 J. Plank, J. Blaha, J. Cordingley, M.E. Wilinska,

L.J. Chassin, C. Morgan, S. Squire, M. Haluzik,

J. Kremen, S. Svacina, W. Toller, A. Plasnik,

M. Ellmerer, R. Hovorka, T.R. Pieber: Multicen-

Part H 76

Automation and Control in Biomedical Systems References 1377

tric, randomized, controlled trial to evaluate blood

glucose control by the model predictive con-

trol algorithm versus routine glucose management

protocols in intensive care unit patients, Diabetes

Care 29, 271–276 (2006)

76.45 C. Cobelli, G. Federspil, G. Pacini, A. Salvan,

C. Scandellari: An integrated mathematical model

of the dynamics of blood glucose and its hormonal

control, Mathem. Biosci. 58,27–60(1982)

76.46 C. Cobelli, A. Ruggeri: Evaluation of por-

tal/peripheral route and of algorithms for insulin

delivery in the closed–loop control of glucose in

diabetes – a modeling study, IEEE Trans. Biomed.

Eng. BME-30, 93–103 (1983)

76.47 C. Cobelli, A. Mari: Validation of mathematical

models of complex endocrine–metabolic systems.

A case study on a model of glucose regulation, Med.

Biol. Eng. Comput. 21, 390–399 (1983)

76.48 E. Salzsieder, G. Albrecht, U. Fischer, E.-J. Freyse:

Kinetic modeling of the glucoregulatory system to

improve insulin therapy, IEEE Trans. Biomed. Eng.

32, 846–855 (1985)

76.49 J.T. Sorensen: A Physiologic Model of Glucose

Metabolism in Man and its Use to Design and As-

sess Improved Insulin Therapies for Diabetes. Ph.D.

Thesis (Department of Chemical Engineering, MIT

1985)

76.50 M. Berger, D. Rodbard: Computer simulation of

plasma insulin and glucose dynamics after subcu-

taneous insulin injection, Diabetes Care 12,725–736

(1989)

76.51 W.R. Puckett: Dynamic Modeling of Diabetes

Mellitus. Ph.D. Thesis (Department of Chemical En-

gineering, University of Wisconsin Madison 1992)

76.52 E.D. Lehmann, T. Deutsch: A physiological model

of glucose–insulin interaction in type 1 diabetes

mellitus, J. Biomed. Eng. 14, 235–242 (1992)

76.53 R.S. Parker, J.H. Ward, N.A. Peppas, F.J. Doyle

III: Robust H

∞

glucose control in diabetes us-

ing a physiological model, AIChE J. 46,2537–2549

(2000)

76.54 P.J. Lenart, R.S. Parker: Modeling exercise effects

in typeI diabetic patients, Proc. 15th IFAC World

Congress on Automatic Control (Barcelona 2002)

76.55 A. Roy, R.S. Parker: Dynamic modeling of free fatty

acids, glucose, and insulin: an extended minimal

model, Diabetes Tech. Theraput. 8, 617–626 (2006)

76.56 A. Roy, R.S. Parker: Dynamic modeling of exer-

cise effects on plasma glucose and insulin levels,

J. Diabetes Sci. Technol. 1, 338–347 (2007)

76.57 DCCT – The Diabetes Control, Complications Trial

Research Group: The effect of intensive treatment

of diabetes on the development and progression

of long-term complications in insulin-dependent

diabetes mellitus, N. Engl. J. Med. 329, 977–986

(1993)

76.58 DCCT – The Diabetes Control, Complications Trial Re-

search Group: The absence of a glycemic threshold

for the development of long-term complications:

The perspective of the diabetes control and com-

plications trial, Diabetes 45, 1289–1298 (1996)

76.59 American Diabetes Association: Continuous sub-

cutaneous insulin infusion (position statement),

Diabetes Care 24(1), S98 (2001)

76.60 N. Hernjak, F.J. Doyle III: Glucose control design

using nonlinearity assessment techniques, AIChE J.

51, 544–554 (2005)

76.61 C.C. Palerm, H. Zisser, W.C. Bevier, L. Jovanovic,

F.J. Doyle III: Prandial insulin dosing using run-

to-run control: application of clinical data and

medical expertise to deﬁne a suitable performance

metric, Diabetes Care 30, 1131–1136 (2007)

76.62 C. Owens, H. Zisser, L. Jovanovic, B. Srinivasan,

D. Bonvin, F.J. Doyle III: Run-to-run control of

blood glucose concentrations for people with Type

1 diabetes mellitus, IEEE Trans. Biomed. Eng. 53,

996–1005 (2006)

76.63 G. Van den Berghe, P. Wouters, F. Weekers: Inten-

sive insulin therapy in critically ill patients, N. Engl.

J. Med. 345, 1359–1367 (2001)

76.64 G.V. Van den Berghe, P.J. Wouters, R. Bouillon,

F. Weekers, C. Verwaest, M. Schetz, D. Vlasselaers,

P. Ferdinande, P. Lauwers: Outcome beneﬁt of in-

tensive insulin therapy in the critically ill: insulin

dose versus glycemic control, Crit. Care Med. 31(2),

359–366 (2003)

76.65 S.J. Finney, C. Zekveld, A. Elia, T.W. Evans: Glucose

control and mortality in critically ill patients, J. Am.

Med. Assoc. 290, 2041–2047 (2003)

76.66 R.S. Parker, E.P. Gatzke, F.J. Doyle III: Advanced

model predictive control (MPC) for type I diabetic

patient blood glucose control, Proc. Am. Control

Conf., Vol. 5 (2000) pp. 3483–3487

76.67 P. Dua, F.J. Doyle III, E.N. Pistikopoulos: Model-

based blood glucose control for Type 1 diabetes via

parametric programming, IEEE Trans. Biomed. Eng.

53, 1478–1491 (2006)

76.68 L.B. Rowell, J.T. Shepherd: Exercise: regulation and

integration of multiple systems. In: Handbook of

Physiology, ed. by L.B. Rowell, J.T. Shepherd (Ox-

ford Univ. Press, New York 1996), Chap. 12

76.69 L.B. Borghouts, H.A. Keizer: Exercise and insulin

sensitivity: a review, Int. J. Sports Med. 21,1–12

(2000)

76.70 S.H. Levitt, J.A. Purdy, C.A. Perez, S. Vijayakumar

(Eds.): Stereotactic radiosurgery and radiotherapy.

In: Technical Basis of Radiation Therapy: Practical

Clinical Applications (2006) p. 244

76.71 J.M. Slingerland, I.F. Tannock: Cell Proliferation

and Cell Death. In: The Basic Science of Oncology,

3rd edn., ed. by J.M. Slingerland, R. Hill (McGraw-

Hill, New York 1998) pp. 134–165

76.72 The American Cancer Society: Cancer facts and ﬁg-

ures: 2008, http://www.cancer.org/downloads/STT/

/2008CAFFﬁnalsecured.pdf (last accessed April 14,

2009)

Part H 76

1378 Part H Automation in Medical and Healthcare Systems

76.73 J.A. Purdy: IMRT, IGRT, SBRT – advances in the

treatment planning and delivery of radiotherapy,

Front. Radiat. Ther. Oncol. 40, 18–39 (2007)

76.74 J.D. Fenwick, S.W. Riley, A.J. Scott: Advances in

intensity-modulated radiotherapy delivery, Med.

Phys. 139, 193–214 (2008)

76.75 M. Langer, E.K. Lee, J.O. Deasy, R.L. Rardin,

J.A. Deye: Oprations research applied to radio-

therapy, and NCI–NSF-sponsored workshop, Int. J.

Radiat. Oncol. Biol. Phys. 57, 762–768 (2003)

76.76 E.K. Lee, T. Fox, I. Crocker: Simultaneous beam

geometry and intensity map optimization in

intensity-modulated radiation therapy, Int. J. Ra-

diat. Oncol. Biol. Phys. 64, 301–320 (2006)

76.77 B. Chachuat, A.B. Singer, P.I. Barton: Global meth-

ods for dynamic optimization and mixed-integer

dynamic optimization, Ind. Eng. Chem. Res. 45,

8373–8392 (2006)

76.78 L.T. Biegler, I.E. Grossmann: Retrospective on opti-

mization, Comput. Chem. Eng. 28, 1169–1192 (2004)

76.79 L. Norton: A Gompertzian model of human breast

cancer growth, Cancer Res. 48, 7067–7071 (1988)

76.80 G.W. Swan: Tumor growth models and cancer

chemotherapy. In: Cancer Modeling,Vol.83,ed.by

J.R. Thompson, B. Brown (Marcel Dekker, New York

1987) pp. 91–179

76.81 A. Asachenkov, G. Marchuk, R. Mohler, S. Zuev:

Disease Dynamics (Birkhäuser, Boston 1994)

76.82 H.S. Friedman, S. Keir, A.E. Pegg, P.J. Houghton,

O.M. Colvin, R.C. Moschel, D.D. Bigner, M.E. Dolan:

O6-bg mediated enhancement of chemotherapy,

Mol. Cancer Ther. 1, 943–948 (2002)

76.83 R.B. Martin: Optimal control drug scheduling of

cancer chemotherapy, Automatica 28, 1113–1123

(1992)

76.84 U. Ledzewicz, U. Brown, H. Schättler: Comparison of

optimal controls for a model in cancer chemother-

apy with L-1- and L-2-type objectives, Optim.

Methods Softw. 19, 339–350 (2004)

76.85 U. Ledzewicz, H. Schättler: Optimal controls for

a model with pharmacokinetics maximizing bone

marrow in cancer chemotherapy, Math. Biosci.

206, 320–342 (2007)

76.86 J.M. Harrold, J.L. Eiseman, W.C. Zamboni,

R.S. Parker: Modeling of toxicity effects and non-

linearities in pharmacokinetics of 9-nitrocampto-

thecin in mice, AIChE Annual Meeting (Austin 2004)

76.87 J.M. Harrold, R.S. Parker: Clinically-relevant cancer

chemotherapy treatment scheduling using param-

eterized mixed-integer programming, Proc. CPC VII

(Lake Louise 2006), CACHE Corporation

76.88 P. Dua, V. Dua, E.N. Pistikopoulos: Optimal deliv-

ery of chemotherapeutic agents in cancer, Comput.

Chem. Eng. 32, 99–107 (2008)

76.89 A. Pertsinidis, I.E. Grossmann, G.J. McRae: Para-

metric optimization of MILP programs and a frame-

work for the parametric optimization of MINLPs,

Comput. Chem. Eng. Suppl. 22

, S205–S212 (1998)

76.90 I.E. Grossman: Review of nonlinear mixed-integer

and disjunctive programming techniques for pro-

cess systems engineering, http://egon.cheme.cmu.

edu/Papers/GrossmannReviewNon.pdf

(last accessed April 14, 2009)

76.91 M. Morari, A. Gentilini: Challenges and opportuni-

ties in process control: biomedical processes, AIChE

J. 47, 2140–2143 (2001)

76.92 M. Morari, E. Zaﬁriou: Robust Process Control

(Prentice-Hall, Englewood Cliffs 1989)

76.93 H. Tähti, H. Nevala, T. Toimela: Reﬁning in vitro

neurotoxicity testing – the development of blood-

brain barrier models, Altern. Lab. Anim. 31, 273–276

(2003)

76.94 S.F. Khattak, K.-S. Chin, S.R. Bhatia, S.C. Roberts:

Enhancing oxygen tension and cellular function in

alginate cell encapsulation devices through the use

of perﬂuorocarbons, Biotechnol. Bioeng. 96,156–

166 (2006)

76.95 R. Gebhardt: Metabolic zonation of the liver:

Regulation and implications for liver function,

Pharmacol. Ther. 53, 275–354 (1992)

76.96 International Life Sciences Institute: Physiological

parameter values for PBPK models, (1994)

76.97 G.B.West,J.H.Brown,B.J.Enquist:Ageneral

model for the origin of allometric scaling laws in

biology, Science 276, 122–126 (1997)

76.98 G.B. West, J.H. Brown, B.J. Enquist: The fourth

dimension of life: Fractal geometry and allomet-

ric scaling of organisms, Science 284, 1677–1679

(1999)

76.99 P. Vicini, M.R. Gastonguay, D.M. Foster: Model-

based approaches to biomarker discovery and

evaluation: a multidisciplinary integrated review,

Crit. Rev. Biomed. Eng. 30, 379–418 (2002)

76.100 M.E. Csete, J.C. Doyle: Reverse engineering of bio-

logical complexity, Science 295, 1664–1669 (2002)

76.101 R.Tanaka,M.Csete,J.Doyle:Highlyoptimised

global organisation of metabolic networks, IEE

Proc. Syst. Biol. 152, 179–184 (2005)

Part H 76

1379

Automation i

77. Automation in Hospitals and Healthcare

Brandon Savage

Healthcare is a complex industry, but yet it lags be-

hind virtually all others in automation and use of

information technology (IT). For healthcare, tech-

nology serves as an untapped catalyst for higher

efﬁciency, lower cost and broader access to care.

The appropriate application can minimize medi-

cal errors, promote better management of chronic

illness, and enable clinicians to intervene earlier

and anticipate prognosis. Additionally, medical

informatics provides the tools to generate new

insights about both individuals and entire pop-

ulations through data analysis and visualization.

These systems can also be used in support of con-

tinuous quality improvement efforts as well as

to reduce inefﬁciencies. While signiﬁcant strides

have been made in the implementation of medi-

cal informatics, there are numerous challenges to

resolve before we will be able to realize the full

beneﬁts of healthcare IT.

77.1 The Need for Automation

in Healthcare .......................................1380

77.2 The Role of Medical Informatics .............1382

77.2.1 Information Management .............1382

77.2.2 Knowledge Management ..............1383

77.2.3 Knowledge Representation............1384

77.2.4 Knowledge Generation .................1385

77.2.5 Knowledge in Action ....................1387

77.3 Applications .........................................1389

77.3.1 Patient Access..............................1389

77.3.2 Healthcare Billing ........................1389

77.3.3 Healthcare Administration ............1390

77.3.4 Clinical Care.................................1391

77.3.5 Patient Connectivity .....................1393

77.3.6 Interoperability............................1394

77.3.7 Enterprise Systems .......................1395

77.4 Conclusion ...........................................1396

References ..................................................1396

Automation is so ubiquitous in our daily lives that few

of us give it a second thought. Whether the automated

teller machine (ATM), a barcode scanner, or wireless

communication, we embed automation as a fundamen-

tal mechanism of our daily lives.

It’s stunningto notice how fewparallel examples ex-

ist in the average hospital or physician ofﬁce, when the

exchange of accurate information can literally be a mat-

ter of life and death. In today’s healthcare system the

majority of medical records are still kept on paper – and

most of them,including prescriptions, arestill handwrit-

ten. Even most electronic are commonly in stand-alone

systems, and unlike ATM s, are not interoperable and

cannot share information.

As an example, if you live in Duluth and need to

check your bank balance while visiting New York City,

it’s easily within reach. If you’re ill while in the Big

Apple and are taken to a hospital emergency depart-

ment, it will be virtually impossible for the medical

staff to check your medical record for allergies and past

history. And, only a few healthcare institutions use bar-

code scanning to conﬁrm accurate dose and treatment

administration.

Managing the complexity of healthcare warrants the

need for automation that same complexity is also the

reason effective automation is so difﬁcult. There are

thousands of data and decision points in the course of

treatment. As medical knowledge expands at a break-

neck pace, what we deﬁne as today’s best practice and

standard protocol, may not exist tomorrow.

Part H 77

1380 Part H Automation in Medical and Healthcare Systems

77.1 The Need for Automation in Healthcare

While informationtechnology has transformed other in-

dustries, the healthcare industry (especially in the US)

remains largely paper-based. This may have been suf-

ﬁcient in the days of the family physician who made

house calls, but is inadequate to meet the needs of

today’s complex healthcare environment. Medicare pa-

tients, for example, see an average of three different

providers regularly – but without a digitized record the

clinicians often lack a complete picture of any patient’s

medical history. Additionally, medical information col-

lected in the physician’s ofﬁce is stored separately from

records related to hospital admissions.

The impacts on the patient and the healthcare

system range from the merely inconvenient to the po-

tentially (and all too often) fatal. Without access to

a complete medical record, clinicians may inadvertently

prescribe medication to which the patient is allergic, or

one clinician may unwittingly prescribe a drug that is

harmful when taken with another medication prescribed

by a different practitioner.

Even when records do exist in electronic form, the

information cannot be readily shared because the sys-

tems are not interoperable. Within a single institution,

interfaces may be required just to move data between

clinical and ﬁnancial information systems, for example,

or between the system that holds pharmacy data and the

one in which nurses document a patient’s vital signs.

Beyond sharing between information systems, shar-

ing information among healthcare organizations is even

Fig. 77.1 Example of healthcare visualization and monitoring au-

tomation to prevent medical errors

more complex. Thereis no single standardfor how med-

ical information is represented electronically, so data

that exists in one vendor’s system cannot be used by

a different vendor’s system. Overcoming the barriers to

interoperability will have a signiﬁcant impact in accel-

erating the adoption of electronic medical records.

The discussion of interoperabilitybegs the question,

however, What are the real beneﬁts of automation in

healthcare? In fact, the beneﬁts are considerable across

the entire spectrum of care.

Consider patient safety. Every year, between 50000

and 100000 people die in the US as the result of pre-

ventable medical errors [77.1]. Countless others suffer

less severe injuries. Errors arise in many different ways,

from prescribing the wrong medication to performing

surgery at the wrong site. Confusion arises when two

different patients have similar names, or when patients

are moved from one hospital room to another. A mis-

placed decimal point can result in a patient getting less

than a therapeutic dose of a drug, or a tenfold overdose:

•

Incorporating alerts into electronic medical records

can help ensure the completeness and accuracy of

the record. The system can remind clinicians to ﬁll

in missing data such as allergy information, or in

the case of surgery which side of the body is to be

operated on.

•

Automating the medication-ordering process can

eliminate hazards that arise from illegible handwrit-

ing; in hospitals, it can shave hours off the time

between when a drug is ordered to when it is ad-

ministered. When expert rules are incorporated, the

system can automatically suggest appropriate medi-

cations based on the patient’s current condition, past

medical history, age and weight, and even the list of

drugs that will be paid for by the patient’s insurance.

•

In addition to electronic ordering, closed-loop med-

ication management automates both the dispensing

and administration of drugs to ensure that the ﬁve

rights are met – i.e., that the right patient gets the

right dose of the right drug via the right route at the

right time. Nurses scan barcodes on each unit dose

of medication and on a hospital patient’s wristband;

if there is a mismatch, the system generates an alert

that can prevent the error.

•

When the drug Vioxx was recalled, most physician

ofﬁces had to resort to reviewing every single ﬁle by

hand to determine which patients were taking the

drug, resulting in delays of days or weeks before

Part H 77.1

Automation in Hospitals and Healthcare 77.1 The Need for Automation in Healthcare 1381

patients could be notiﬁed of the recall and given an

alternative prescription. Clinics that had electronic

record systems,on the other hand, wereable to com-

pile lists of which patients were taking Vioxx and

generate notices to their primary care physicians

within a matter of hours.

In the treatment of chronic illness, a handful includ-

ing cardiovascular disease, diabetes, and asthma, are

responsible for the lion’s share of medical expenses.

Yet, if these conditions are managed appropriately,

much of the cost can be avoided and patients can con-

tinue to lead long and productive lives without acute

ﬂare-ups of the disease. Electronic record systems can

automatically generate alerts and reminders to help en-

sure that patients receive effective preventive care, such

as annual eye exams for diabetic patients.

In addition, being able to run outcomes analysis on

all patients with a certain condition can help identify

trends and determine what treatments are more effec-

tive than others within a certain population. This kind

of analysis, like drug recall notiﬁcation, is simply not

practicable with paper records, but can be done rela-

tively easily with the right kind of electronic records

systems.

Increased Efﬁciency and Accuracy. At each sep-

arate encounter with the healthcare system, patients

are usually asked to provide the same basic informa-

tion including name, date of birth, and insurance. Each

time the data is taken down presents an opportunity for

transcription error, as well as a frequent source of an-

noyance for patients. Automated check-in kiosks can let

patients review their demographic data on ﬁle and make

any needed changes, without having to recite the same

litany over and over again.

Nurses spend as much as one third of their time

on charting and other administrative tasks. Automation

can reduce the amount of time nurses spend on docu-

mentation, freeing up more time that can be spent on

direct patient care. Clinicians can pull patient informa-

tion from one part of the system to another, so it does

not have to be re-entered.Nurses can chart byexception,

entering only data that is outside normal ranges or that

has changed since the last reading.

Unlike paper records, electronic records give all

members of the care team simultaneous access to a pa-

tient’s most up-to-date clinical information. Lab results,

for example, are available online as soon as they are

completed, avoiding redundant tests. Physicians can

check on their hospitalized patients from ofﬁce or home

at any time and enter new orders, rather than waiting for

daily rounds.

Considering consistency of care, the quality of

healthcare varies enormously from one hospital to an-

other, and even within the same hospital from one

provider to another. Simple best practices that are

proven to save lives – such as presurgical antibiotics

or administering beta-blockers to heart attack patients

within 24h – are not uniformly followed, even within

a single institution [77.2]. Even with recommended care

guidelines and best practices published for numerous

disease processes, Americans as a whole receive only

54.9% of the recommended care [77.3], and fewer than

half of American children receive timely, needed med-

ical care [77.4]. Additionally, 11% of care received is

either not recommended or potentially harmful [77.3].

In large part, this is due to the fact that the already

overwhelming body of medical science is growing at

an explosive pace, and no individual can keep up with

it. Clinicians, like the rest of us, can also be resistant to

change – preferring to continue using the same protocol

they always have for a certain condition. To promote

higher consistency, healthcare institutions can develop

standard order sets that are used for speciﬁc conditions.

While some hospitals use hard-copy versions of such

order sets, digitizing them and integrating them with

a clinical information system, so they are automatically

presented at the point of care can be more effective

consistency.

On average, it takes 17 years for a new medical dis-

covery to become part of the standard of care. Clinical

information systems can accelerate this process through

the use of expert systems that are integrated with the

electronic medical record to make recommendations

about appropriate treatment and present that informa-

tion to the clinician at the point of care. Evidence-based

practice (EBP) is an approach to healthcare that effec-

tively integrates best practices with the daily activities

of patient care, resulting in a consistently high standard

of care both within and across healthcare organizations.

EBP can reduce inappropriate variations incare while at

the same time enabling the appropriate tailoring of best

practice to each patient’s clinical condition [77.5].

Currently only a few of the largest healthcare in-

stitutions are able to implement expert systems in this

way; the challenge going forward will be to make

the technology accessible to smaller organizations with

limited IT resources.

Part H 77.1

1382 Part H Automation in Medical and Healthcare Systems

77.2 The Role of Medical Informatics

Since its inception in the 1960s, there has been a steady

increase in demand for medical informatics, includ-

ing dramatic growth occurring in the past decade. The

reason for this explosive growth is in how medical in-

formatics provides a foundation that bridges the art and

science to address challenges facing healthcare. In order

to do this, we ultimately need to move medical infor-

matics beyond the level of information management

– acquiring, storing, and moving data – to knowl-

edge management – the ability to synthesize and apply

available data and thereby transform the practice of

medicine.

The breadth of patient data and how that data

is utilized in a healthcare setting is important in

understanding the distinction between knowledge man-

agement and informatics. There are two categories of

patient data. The ﬁrst is subjective data, such as pain or

shortness of breath, that is qualitatively described by the

patient. The second is objective data, such as heart rate

or weight, which can be veriﬁed by external observers.

Objective data can further be broken down into pa-

tient demographics(name, age,etc.), observations made

by healthcare providers, and quantitative parameters

representing the anatomic, biochemical, and physio-

logic states of the patient. Within the latter category,

advances in medical diagnostics are not only creating

a more comprehensive picture, they are increasingly

shifting from anatomic to physiologic parameters –

i.e., functional as opposed to structural data. A com-

puted tomography (CT) scan, for example, can detect

an anatomic abnormality in the chest, while a positron

emission tomography (PET) scan of the same area can

detect metabolic characteristics that suggest the mass

is a malignant tumor. Speciﬁc biochemical markers are

Fig. 77.2 Healthcare automation has revolutionized the

practice of medicine with innovations to detect and diag-

nose diseases earlier

better indicators of the effects of congestive heart fail-

ure than are imprecise observations of distended veins

in the neck.

Another important shift is from phenotypic data –

the observable signs or symptoms of normal or ab-

normal gene expression – to genotypic data – the

underlying genetic code itself. Not only is genotypic

data more speciﬁc (i. e., there can be many different

genotypes that present the same phenotype) but it can

often predict disease before any outward symptoms are

present. The BRCA-1 gene, for example, presents a sig-

niﬁcantly elevated risk of breast cancer. Women who

have this gene should be more closely monitored so

that cancer is detected early, when it is more easily

treated; some women may opt for mastectomy as a pre-

ventive measure. Genotypes for certain liver enzymes

can predict how fast a patient will metabolize a certain

medication even before that patient takes the ﬁrst dose;

this enables clinicians to more precisely prescribe the

optimum amount of the drug.

77.2.1 Information Management

With this vast array of patient data, information man-

agement is a necessary ﬁrst step in delivering safe,

effective, high-quality healthcare. The following are

four components of an information management strat-

egy:

•

Acquisition – the transformation of observable

objective phenomena and recordable subjective ex-

periences into data, which can then be stored in

paper or digital formats.

•

Collation – the organization of thesedata into a nav-

igable structure linking the data to patients and

populations.

•

Distribution– the movement of information from its

source to potential users of the data.

•

Access – the interface that enables a user to query

data once it has been distributed.

Failure in any one of these components will re-

sult in critical decisions being made based on only

a small subset of available data, with impact both ﬁ-

nancial and physical. It is estimated, for example, that

clinicians order approximately $2.5 billion worth of re-

dundant laboratory tests annually because they do not

have access to previously obtained results [77.6]. The

consequence of administering medication without ac-

cess to complete allergy information can be fatal.

Part H 77.2

Automation in Hospitals and Healthcare 77.2 The Role of Medical Informatics 1383

Clinical workflow

Application logic

Clinical data storage Data analysis

Data warehouse

Clinical data access

Clinical decision support

Information access

Collaboration tools

Documentation Care plans/orders

Config

management

Clinical

content

management

Contolled

medical

vocabulary

Fig. 77.3 Clinical informatics-oriented view: application architecture connects clinical workﬂow and is composed of

information access, documentation, action plans, and collaboration

77.2.2 Knowledge Management

In many other industries, the value of information man-

agement alone has been enough to drive high rates

of IT adoption. This is not the case in healthcare,

however, where adoption of IT hovers around 20%, de-

spite signiﬁcant acknowledged potential beneﬁt. The

complexity and quantity of medical data decreases,

information management as notsufﬁcient to driveadop-

tion. Instead, medical informatics needs to emphasize

knowledge management that focuses on the content and

use of information technology.

Where information management lags in acquiring

and conveying data, knowledge management gains ad-

ditional value from the meaning, context, and structure

associated with that data. For example, consider the

relatively simple act of measuring a patient’s blood

pressure. The systolic pressure reading (the 120 in

a reading of 120/80, for example) measures the point at

which thepeak pressureof the blood in a patient’s artery

exceeds the pressure of the blood pressure device’s

cuff as it slowly deﬂates. The blood ﬂow is detected

either by a clinician using a stethoscope or by a trans-

ducer in the device. The resulting number can then be

recorded, time stamped, and associated with a speciﬁc

patient. This piece of information is useful but limited

– the number alone has little meaning without the con-

text of the data collection: Was the pressure recorded

from the left or right arm? Was it recorded using an

external blood pressure cuff or a transducer within an

artery? Was thepatient lying down or standing?(A read-

ing taken while the patient is standing will always be

lower than one taken when the patient is sitting or

lying down.) Was the observation made by a health pro-

fessional or reported by the patient? All these bits of

information can affect the interpretation and utility of

the information.

In aggregating and visualizing data, such as dis-

playing a graph of a patient’s blood pressure values

over time, an information-driven system is limited in

its ability to select appropriate data points. The system

may include values that are not comparable (combining

both sitting and standing blood pressure, for example)

or may fail to include comparable values that were

Fig. 77.4 The complexity and quantity of medical data has

been a consistent barrier to the adoption of electronic med-

ical records (EMR) and other clinical technologies

Part H 77.2

1384 Part H Automation in Medical and Healthcare Systems

recorded differently(i.e., from asensor inserteddirectly

in the artery rather than from a blood pressure cuff).

A knowledge-driven system, by contrast, could evalu-

ate all possible blood pressure data and include only the

appropriate values in the trending task.

With a knowledge-driven healthcare system, clini-

cians use knowledge management to transform infor-

mation into insight, which in turn drives more effective

healthcare actions. The result is a system that delivers

the highest value (i.e., best quality care) at the lowest

cost.

Knowledge management comprises three types of

knowledge: patient-centered (insight into the available

data about a patient), healthcare process (insight into

effectiveness and value), and medical (body of medical

science and best practices).

There are three key stages of knowledge manage-

ment that allow these three types of knowledge to drive

improvements in the healthcare system: knowledge rep-

resentation, knowledge generation, and knowledge in

action.

77.2.3 Knowledge Representation

The challenge for knowledge representation is to estab-

lish the meaning, context, and structure of data, as well

as to enable the portability of knowledge across tradi-

tional data boundaries – whether software applications,

between different locations within a single healthcare

organization (such as an outpatient clinic and a hospi-

tal), or entirely separate institutions.

In the above example, knowledge representation

would provide context to the blood pressure reading

and corresponding heart rate measurement and whether

the patient is taking any medications that modify blood

pressure (information that may have to cross organi-

zational boundaries). Additionally, the system could

provide other relevant information from the patient’s

medical history and from external references, including

whether the patient is being evaluated for heart dis-

ease (associated with high blood pressure) or a severe

infection (associated with low blood pressure).

Knowledge representation is the result of robust

medical terminologies that form the fabric upon which

electronic medical records can be designed. These

medical terminologies consist of controlled medical

vocabularies (external standardized referenced dictio-

naries that attribute speciﬁc meanings to concepts that

can be linked to data ﬁelds) and medical ontologies that

contain relationships between concepts to establish the

structure of stored data.

In practice, the language of medicine can vary

tremendously fromone institutionto another. Whileone

institution diagnose an episode as a heart attack another

institution may refer to the episode as a myocardial in-

farction. In the mid-1990s, Medicalogic created one of

the ﬁrst electronic medical records that enforced what

was then a controversial design principle: using a cen-

trally managed dictionaryof observational terms that all

customers could access. The result was a knowledge-

based community that allowed healthcare institutions to

create advanced documentation forms that referenced

the dictionary; these forms could also be shared with

other institutions. Themembers ofthe community could

collaborate on innovation as well as aggregate their data

to support the creation of common performance indica-

tors and benchmarks. Vendors also recognized business

opportunities in the creation of other content that could

be distributed across this customer base. For example,

Clinical Content Consultants was founded in 2000 by

John Janas III, M.D., and John R. Thompson, M.D.,

to design and implement interoperable evidence-based

forms for clinical decision support in the management

of chronic diseases [77.7].

The development of medical ontologies took off in

the mid-1990s.The Regenstrief Institute at Indiana Uni-

versity launched LOINC (logical observation identiﬁers

names and codes) to enable the electronic movement

of clinical data from laboratories that produce the data

to hospitals, physician’s ofﬁces, and payers who use

the data for clinical care and management purposes. At

about the same time, the College of American Pathol-

ogists teamed up with Kaiser Permanente to expand

SNOMED (systematized nomenclature of medicine) to

include both a dictionary of medical terms and an on-

tology of relationships for these terms, both of which

were independent ofa speciﬁc electronicmedical record

vendor.

The relationships deﬁned in most established

medical terminologies are primarily parent/child or

synonym-typerelationships. Despite their apparent sim-

plicity, the impact on IT system design is dramatic.

Two core IT needs illustrate the impact of an on-

tological hierarchy on information. The ﬁrst is search.

One of the biggest challenges of an electronic medical

record (EMR) is ﬁnding information quickly, whether

ﬁnding the right medication in a list of hundreds of

thousands of alternatives or ﬁnding the relevant data to

aid in performing a speciﬁc task. Where search engines

such as Google use inherent linkages present within the

data (such as hyperlinks) to infer associations, these

linkages do not appear in the medical record, and med-

Part H 77.2