Nof S.Y. Springer Handbook of Automation

Подождите немного. Документ загружается.

Medical Informatics 80.3 Communication and Integration 1425

and (2) results of observations. Figure 80.1 shows two

computer screens for a medical doctor when checking

patients. The left screen is used to display the results

of observation (x-ray), whereas the right screen is an

order entry system to show anamnesis and other re-

sults of tests. Diagnosis may take a long time, since

the patients maydescribe their complaintsincompletely,

which may lead the doctor to order the wrong tests

or treatments. Then, an iterative observation-diagnosis-

therapy occurs. For patients who have chronic diseases,

iterations may repeat over and over for a long time pe-

riod. However, sometimes the iterations occur because

of mistakes/errors resulting from doctors, pharmacists,

nurses, or inspecting staffs and/or facilities.

Therapy cannot be done without the orders from

doctors. It may be done through medication, surgery, re-

habilitation, or diet control. Technologies of therapy are

constantly evolving. Due to its advancement, life quality

of patients is improving.

Observation, diagnosis, and therapy form a cycle

(Fig.80.2). The cycle stops only when the patient is

cured or decides to quit the cycle.

80.3 Communication and Integration

Medical services in modern hospitals are fulﬁlled by

a group of specialists: physicians, nurses, pharmacists,

biologists, and administrative and laboratory staffs.

Moreover, medical services are fulﬁlled by advanced

facilities, and information and computer technologies.

Patients are cared for by an integrated system consisting

of the above-mentioned specialists and expensive facil-

ities. The integrated system can function efﬁciently and

effectively only when theindividualpersons or facilities

can be integrated through a communication system.

Communication as a key component for integra-

tion requires two elements: channels and interfaces.

Channels mean how the persons or facilities are

connected. Most medical facilities can offer commu-

nication with a local area network (LAN) by applying

TCP/IP standard protocol. Therefore, specialistscan up-

load/download or operate the facilities within the LAN.

Associated clinics or outside laboratories may be autho-

rized to use the Internet as a channel to obtain data from

the facilities of hospitals. For short distance transmis-

sions, WiFi provides a wireless alternative of communi-

cation. It is useful for a doctor to display patient records

or to give a medical order when checking with a patient

in a ward with a mobile/wireless handheld device, per-

sonal digital assistance (PDA) for example. Figure 80.3

shows a portable computerized order entry device. It is

a WiFi and web ready system. By its powerful com-

puting, communication, and graphic capabilities, physi-

cians can display the test results of patients and give

treatment or medication orders when they are checking

with patients anywhere in the WiFi coverage zone.

Interface means that facilities speak the same lan-

guage. Usually an interface is fulﬁlled by standards.

For administrative data exchange in healthcare, Elec-

tronic Data Interchange for Administration, Commerce

and Transport (EDIFACT) is used in Europe [80.3],

whereas Health Level 7 (HL7) is used in the United

States [80.4]. With the standards, participants (such as

clinics, laboratories, drug stores, and insurance com-

panies) can encode and decode messages by the same

format, which speeds up communication and enables

integration of processes occurring between units of

a hospital or between hospitals. For medical images

such as x-ray, CT, and magnetic resonance imag-

ing (MRI), digital imaging and communication in

medicine (DICOM) is a standard for image communi-

cation [80.5]. DICOM also plays a role of standardizing

the image inputs/outputs of medical facilities developed

by various vendors.

Fig. 80.3 Portable medical order-entry device (photo cour-

tesy of DVNET Technology Co., www.dvnet.biz)

Part H 80.3

1426 Part H Automation in Medical and Healthcare Systems

80.4 Database and Data Warehouse

Most healthcare organizations host large databases to

maintain the demographical records, complaints and

diagnostic records, laboratory tests and inspections, or-

ders of therapy, and insurance and payment records that

are associated with each patient. The massive amount

of healthcare information requires a database that has

a storage capacity of multiple terabytes (TB). There

are several reasons why a large storage capacity is re-

quired. First, voice, image, and video types of test and

inspection data are gradually playing an important role

in healthcare. They usually take up large data storage

space. Besides, health authorities of government and

insurance companies usually demand hospitals to main-

tain the records for a certain period, which also results

in the need for large storage space. Although the cost

per byte of data storage is decreasing yearly, larger and

more urgent demands on storage space still increase the

total cost of database system year by year in modern

healthcare organizations.

Though today’s object-oriented database (OODB)

seems to be a suitable solution to accommodate var-

ious data formats, which include text, audio, video,

animation, and image, and can manage linked rela-

tionships easily, its low performance in transactions

hinders the applications of OODB in healthcare organi-

zations. Relational database (RDB) presenting database

in ﬂat tables is a typical database for healthcare orga-

nizations. The advantages of RDB include supporting

the standard structural query language (SQL), scalable

Time

time_key

day

day_of_the_week

month

quarter

year

Drug

drug_key

item_name

brand

type

supplier_key

Address

address_key

street

city

province_or_street

country

Supplier

supplier_key

supplier_name

adress_key

supplier_type

time_key

drug_key

doctor_key

address_key

ID_key

dollars_sold

order_key

Patient order table

time_key

drug_key

supplier_key

dollars_cost

Purchase fact table

Doctor

doctor_key

doctor_name

department

Demography

ID_key

gender

date_of_birth

phone

marriage

Fig. 80.4 A fact constellation scheme of a data warehouse for patient records

performance, and easy maintenance. Besides, because

of the abundance of human resource supply in the

ﬁeld of RDB development, operation, and mainte-

nance, RDB is commonly used in healthcare database

applications.

However, maintaining large and complicated med-

ical databases may become a ﬁnancial burden for

healthcare organizations, unless the value of medical

databases is created. To utilize the value of a large

medical database relies on the development of a data

warehouse. Data warehouse is a collection of associated

databases. Databases in a data warehouse are subject-

oriented and nonvolatile. Subject-oriented means that

the database storage focuses on a special issue, such

as patient treatments or medical records for certain

cancers. Nonvolatile means that the data in a data ware-

house should be extracted from the databases and are

not changed by online transactions. Besides, the online

analytical process (OLAP) is an important function to

retrieve useful information from a data warehouse. Ba-

sic command functions of OLAP include rolling up,

drilling down, slicing, dicing, etc. [80.6]. However, to

successfully apply OLAP depends on how well the data

warehouse is deﬁned.

A data warehouse is deﬁned by specifying the link

relationships among ﬂat tables. Figure 80.4 shows an

example to demonstrate link relationships among tables

in a data warehouse with fact constellation. It shows

that there are many keys for patient order tables. Each

Part H 80.4

Medical Informatics 80.5 Medical Support Systems 1427

key is linked to other tables. By the links, a user can

give an OLAP command. For example, showing the

total patient orders for the last month (dicing through

time

key), then showing the patients according to the

departments they went to visit (drilling down). The

drilling down command is possible because the doc-

tor

key links both the patient order table and the doctor

table. Through the link, each patient order can ﬁnd the

doctor

key. By querying in the doctor table with the

doctor

key, the department is found.

80.5 Medical Support Systems

Today’s healthcare activities are supported by informa-

tion systems and digitized facilities. For this reason,

care of patients has become more efﬁcient, timely, and

accurate. Medical support systems, besides the hospital

information system (HIS) that manages both informa-

tion of patient care and administration, includes at least

four subsystems:

1. Solitary imaging information system

2. Laboratory information system

3. Functions of hospital pharmacy

4. Nursing information systems.

In the following sections, the four subsystems are intro-

duced.

80.5.1 Solitary Imaging Information

Systems

Most modern hospitals own expensive solitary imag-

ing systems, such as x-ray, CT, MRI, digital subtraction

angiography (DSA), gamma cameras, and positron

emission tomography (PET). CT, MRI,andDSA are

mostly found in the radiology department, whereas

gamma cameras and PETs are found in the nuclear

Fig. 80.5 A multislice CT volume scan system (photo

courtesy of Cheng Ching General Hospital)

medicine department, since with these radioactive ma-

terial, which is under government management, is

injected into the veins of a patient in the inspection

process. Figure 80.5 shows a multislice CT volume

scan system. CT is a typical noninvasive examination

in today’s healthcare services. An examination on a CT

involves various kinds of monitor and control of the

CT system. Figure 80.6 shows multiple monitoring and

control functions onboth patientsand the CT system. To

integratethe solitary imaging systems with HIS requires

a standard for exchanging image data. As mentioned

in Sect.80.3, DICOM is a standard for image data ex-

change. With it, associated healthcare practitioners can

view the images across different systems, whether they

be inside or outside the hospital. An example of a CT

image is shown in Fig. 80.7. Some radiologists even

check images of patients in various hospitals through

the Internet.

80.5.2 Laboratory Information Systems

Besides solitary imaging systems, many tests may be

performed in a laboratory or even a clinical room

by portable devices. Equipment in a laboratory may

Fig. 80.6 Multiple monitoring and control functions on

both patients and CT system

Part H 80.5

1428 Part H Automation in Medical and Healthcare Systems

Fig. 80.7 Example of a CT image (photo courtesy of

Cheng Ching General Hospital)

include an electrocardiogram (ECG), ultrasound, or

spirography. Biosignals produced by the equipment are

digitalized and conformed to DICOM, so that they can

be integrated into a HIS. Additionally, those biosignals

are collected synchronously with patients, which means

that the patients must stand by when the biosignals are

recording.

Besides the previous synchronous examination,

some examinations are asynchronous. In other words,

the examinations only need specimens from patients

and do not need patients to wait while the equip-

ment is examining. Hospitals host laboratories to give

asynchronous examinations of specimens from patients,

such as blood and urine. Electron microscopes and

centrifuges are typical examples of equipment for lab-

oratories. To perform pathological examinations, the

systematized nomenclature of medicine (SNOMED) is

a code system that indicates various aspects of a dis-

Fig. 80.8 Examples of specimens (each with a barcode)

ease [80.7]. It is a checkbook for diseases based on

specimens from patients.

There are two basic activities that a hospital labora-

tory has to do for either synchronous or asynchronous

examinations. The ﬁrst activity is about getting sam-

ples or biosignals from patients. In the ﬁrst activity,

there must be a system for registering the patient ID

and the examinations that were ordered by a doctor.

For an asynchronous examination, usually a specimen

from a patient is labeled by a barcode that shows the

identical patient ID to avoid mismatching with other

patients’ specimens (Fig.80.8). As long as the speci-

mens’ barcodes are recorded into HIS, laboratory staff

schedule a time for the examination. The second activ-

ity is to deliver the examination report to HIS. For both

synchronous and asynchronous examinations, the ex-

amination reports are presented in text, audio, or video

and have to be stored in the database of HIS so further

diagnoses or therapies can be ordered by a doctor.

Usually a laboratory information system involves

various kinds of workﬂows. Those workﬂows instruct

the administrative and laboratory staffs to efﬁciently

deliver services to patients.

80.5.3 Hospital Pharmacy Information

System

The hospital pharmacy is an important service unit that

supports care of patients. Usuallythe organization of the

hospital pharmacy is divided into two parts, one for in-

patients and the other for outpatients. The information

system of the hospital pharmacy includes an inventory

planning and control system that monitors the inventory

levels of drugs and purchases drugs whenever neces-

sary to maintain high service quality and low inventory

cost of the pharmacy. Besides, the pharmacy informa-

tion system for outpatients should maintain and manage

the dispensing information given by clinicians. How-

ever, it should allow immediate changes if some other

concerns such as drug interactions occur. For inpatients,

the information system should monitor the drug usage

of patients, so further therapeutic and billing procedures

can proceed based upon this information.

A more advanced pharmacy information system has

an embedded knowledge base system todetect the prob-

lems of drug interactions and mistaken orders issued

by doctors. The pharmacy information system is usu-

ally a subsystem of an HIS in modern hospitals. Some

hospitals may not maintain a hospital pharmacy for

outpatients. However, the dispensing information is de-

livered through the Internet to drug stores designated

Part H 80.5

Medical Informatics 80.6 Medical Knowledge and Decision Support System 1429

by outpatients. In this case, an HL7 or EDIFACT proto-

col will be needed to provide a standard communication

between hospitals and drug stores [80.8]. The current

version of HL7 relies on an XML based message syn-

tax. Through XML syntax, a doctor can embed required

drug dispensing information, e.g., patient, drug product,

dosage, etc., into a message, which is then delivered to

the designated drug store.

80.5.4 Nursing Information Systems

According to Murnane et al. [80.9], three key nurs-

ing information modules are required to successfully

deliver nursing care to patients:

1. Care plan

2. Workload allocation

3. Resource scheduling/monitoring.

HIS provides general patient information, and treat-

ment and drug orders from physicians. However, the

information is not enough for nursing care as soon as

a patient enters into a ward or intensive care unit. To de-

tail the nursing care and utilize the nursing human and

equipment resources relies on an integrated nursing in-

formation system that can seamlessly interact with HIS.

A nursing information system can indicate the relation-

ships among ward, patients, nurse, working shifts, and

basic nursing requirements of the patient.

80.6 Medical Knowledge and Decision Support System

MYCIN is an early yet typical decision support system

that adopts expert systems, or now called knowledge

base systems, in medicine [80.10]. It was designed to

diagnose infectious blood diseases, and then to give

recommendation on the use of antibiotics. There are

only few medical cases like MYCIN in today’s medical

applications. The reason is that to make a knowledge

base useful relies on complete and correct knowl-

edge of medicine. Partial or incomplete knowledge

may lead to an incorrect decision, which may guide

a doctor to a wrong decision on patient care. Besides,

because of the advancing medical knowledge and tech-

nology, to keep a medical knowledge base system up to

date would be a huge burden for medical doctors and

computer specialists in debugging the decision rules.

However, medical knowledge and management has im-

proved by information technologies in two aspects:

evidence-based medicine and data mining. These two

aspects are addressed especially in the following two

sections.

80.6.1 Evidence-Based Medicine

Evidence-based medicine (EBM) is deﬁned as [80.11]:

The conscientious, explicit and judicious use of cur-

rent best evidence in making decisions about the

care of individual patients, based on an integra-

tion of individual clinical expertise with the best

available external clinical evidence from systematic

research and patient’s unique values and circum-

stances.

Fundamentals of EBM include three major steps:

1. Knowledge ﬁndings based on medical evidence

databases

2. Applications of the ﬁndings to patients

3. Evaluating the applications and modifying the

knowledge ﬁndings if necessary.

Knowledge ﬁndings usually involve many database

and statistical operations on the medical data. Most

healthcare journals play a role in publicizing medical

treatments and knowledge ﬁndings for medical doctors

and scientists. Unlike traditional healthcare research

that usually involves limited number of cases, by the

introduction of digital databases and government sup-

port for accumulating patient and treatment records, the

number of cases involved in the study based on the

government support dataset (e.g., the Surveillance, Epi-

demiology, and End Results (SEER) program [80.12]

and NHIRD of Taiwan [80.13]) is very large. The num-

ber of cases can be as large as one million or more.

Thus, the reliability of the research results increases,

which implies that the reliability of such an EBM is also

higher.

80.6.2 Data Mining Techniques

Large-scale medical databases not only increase the re-

liability of EBM research results, they lead to other

potential uses of the data. Traditional EBM studies

usually applied statistical tools only. By data mining

techniques [80.6], medical researchers are able to ﬁnd

other evidence that usually involve more variables and

Part H 80.6

1430 Part H Automation in Medical and Healthcare Systems

are not based ontesting the hypotheses made in advance

by people. Notwithstanding, data mining techniques

can be applied to investigate medical improvement and

patient relationship management in general hospital

management based on the records of patient visits.

Major data mining tools that may help healthcare

administration or healthcare services include:

1. Classiﬁcation tools, e.g., decision trees and neural

networks

2. Clustering techniques, e.g., K-Means and hierarchi-

cal clustering

3. Association rules, e.g., apropri algorithm and rough

sets theory.

Classiﬁcation is a tool to identify the relationships

between certain conditional (independent) variables

and a decision (dependent) variable. The relationships

are mostly speciﬁed to a certain range of conditional

variables (e.g., age between 20 and 25 or systolic hyper-

tension above140mmHg), and certainrange ofdecision

variables (e.g., annual cost of medical care between

$ 10000 and $ 15000). Because tools of decision trees,

such as ID3 and CHAID, provide clear IF-THEN rules

for presenting the relationships between variables, they

are more acceptable, in comparison with neural net-

works that deﬁne variable relationships as weights on

connections of neurons.

Clustering is a tool to separate a data set into

multiple subsets. A predeﬁned distance that measures

similarity between two cases and methods for ﬁnding

clusters (subsets of data) are the two key elements in

clustering. Clusteringdoes not require a dependent vari-

able. Usually it is applied to handle data with a massive

number of variables. Clustering methods (algorithms)

iteratively check the distance between cases to ﬁnd

clusters. Basically, the distance between two cases that

belong to the same cluster should be shorter than the

distance between two cases that belong to two different

clusters. The results of clusters are usually difﬁcult to

interpret and apply, since distance is an unusual con-

cept in medicine, and the cases of each cluster may

include a large number of variables. Even so, cluster-

ing has been applied as a technique for analyzing gene

expression data in the biological ﬁeld, e.g., [80.14], and

problems of clinical and medical care, e.g., [80.15]. For

example, Jannin and Morandi successfully apply deci-

sion tree and clusteringtechniques to predict parts of the

surgical procedure forbrain-tumor patients based on the

pathology-related characteristics of the patient [80.16].

Tools of association rules intend to ﬁnd causal-

ity between variables. For example, patients who have

diabetes are also hypertensive (15%, 90%). The 15%

represents support for such a rule, whereas the 90% rep-

resents conﬁdence. Support deﬁnes the proportion of

the number of patients who have both diabetes and hy-

pertension over the totalnumber of patients.Conﬁdence

deﬁnes the proportion of the number of patients who

have both diabetes and hypertension over the number

of patients who have diabetes. An advantage of associ-

ation rules is that they can identify a causality that has

less support but high conﬁdence. To search for a plau-

sible curable treatment or medication for rare disorder

patients is a typical problem that can apply techniques

of association rules, for two reasons. First, rare disorder

patients mean that the support is small. Second, if the

majority of the patients are cured by a common treat-

ment or medication, its conﬁdence is high. Traditional

statistical tools that mostly do not allow a large number

of variables in the analysis and count on a large number

of samples are not able to handle such a problem.

Though data mining provides an opportunity for

ﬁnding medical evidence that may not be found by tra-

ditional statistical tools, few applications are performed

in the healthcare ﬁelds. Some researchers ascribe the

reasons to a large number of false alarms created by

data mining. Actually, the credibility of mining results

can be proved by cross-validation or some other re-

lated tools [80.17]. From the author’s perspective, there

are two major obstacles for data mining applications in

healthcare. First, healthcare practitioners may not be so

familiar with data warehouse and data mining. Second,

many legacy medical data tables are not normalized. It

is believed that both obstacles will be overcome in the

near future, because many multidisciplinary research

opportunities have been created for both healthcare and

computer science professionals.

80.7 Developing a Healthcare Information System

To develop a good healthcare information system re-

lies on at least three elements: the right people, the

right project management, and the right strategic plan.

Right people means that the healthcare information sys-

tem should cover the operation needs of major system

users (e.g., administrators, physicians, nurses, staff, pa-

Part H 80.7

Medical Informatics 80.8 Emerging Issues 1431

tients, etc.) Questions like how many people and what

kind of people should be included in the information

system depend on the budget and strategic plans of the

healthcare organization. The development of a health-

care information system is a complicated project. It

usually includes stages of requirement analysis, system

design or introduction, system implementation, system

tests, user training, and system validation. The range of

the project duration depends on the type of informa-

tion and may be one month to one year. Besides, the

people involved in each stage vary. Therefore, a well

organized project management is required to ensure

that the project can be successfully implemented and

applied in accordance with the planned schedule and

budgets.

The strategic plan for the healthcare information

system has to be set right so it can match the orga-

nization’s goals and external and internal environment

analysis. Because today’s HIS plays an important role

in patient care and costs much in investment, its success

depends on how well it meets the healthcare organiza-

tion’s goal. The steps of the development of a strategic

plan for a healthcare information system can be summa-

rized as follows:

1. Perform SWOT (strengths, weaknesses, opportuni-

ties, and threats) analysis for the organization.

2. Identify speciﬁcally how information systems can

strengthen the organization’s capabilities to catch

external opportunity.

3. Identify speciﬁcally how an information sys-

tem can avoid threats from internal and external

environments.

4. Identify healthcare information goals of the

organization.

5. Evaluatethe gap betweenthe healthcareinformation

goals and current information system.

6. Identify long-term information system objective

such that the gap can be ﬁlled.

7. Identify resource requirements to fulﬁll the

objective.

8. Make a budget plan for the resource requirements.

80.8 Emerging Issues

The advent of modern medical information technolo-

gies has largely improved patient care and the quality

of care. More speciﬁcally, a healthcare organization al-

most cannotfunction without the support of information

systems. However, there are still many research issues

waiting investigation. Some of them are addressed in

the following sections.

80.8.1 Quality of Care

How to utilize information technologies to improve the

quality of cares is always an issue. Each time a new

technology is introduced, the possibility for improve-

ment of care quality evolves. By means of today’s

radio frequency identiﬁcation (RFID) technology, er-

rors in drug dispensing and operation rooms have been

reduced because an RFID reader system provides an

active alarm when a person, an article, a paper order,

a medicine, or a specimen embedded with an RFID

tag is not in the right place, right position, and right

time that he/she/it is supposed to be. Some case stud-

ies even suggested using RFID to monitor the walking

routes through wards of medical doctors and nurses,

so an alert system may be built to control nosoco-

mial infections. However, side effects of introducing

RFID system are inevitable, just like the side effects

of introducing barcode systems into the healthcare ser-

vices [80.18].

80.8.2 Security

Today’s healthcare organizations carry a lot of infor-

mation about patients. Securing the digital data today

is as important as securing the paper patient records

in the past. Hackers and computer viruses are the two

major sources that may jeopardize the safety of patient

records. It becomes ﬁnancial and technical burdens for

healthcare organizations to secure large patient-record

databases. Some hospitals outsource healthcare infor-

mation systems to IT companies, so they can ease the

technology burden. However, ethical and legal issues

may arise if patient records are lost or stolen from those

outsourcing IT companies, for any reason.

80.8.3 Public-Use Databases

for Medical Research

To attract researchers to get involved in medical

studies, many governments are systemically building

public-use database for medical research. SEER is

Part H 80.8

1432 Part H Automation in Medical and Healthcare Systems

a typical database that accumulates cancer patients’

records (including demographical, histological, and

pathological information). The Taiwan government also

allows researchers to retrieve patient records (including

demographical information, insurance claims, drug dis-

pensing, and hospital information) for medical studies.

Such public-use databases give great opportunities for

researchers to investigate prevalence, survival analysis,

and other medical comparative studies. Since such stud-

ies are performed based on a large number of samples,

they can avoid errors resulting from a limited number of

samples ordata sources.Additionally, more data mining

technologies are applied to the databases to handle their

multiple dimension data tables.

References

80.1 N.B. Sarter, D.D. Woods, C.E. Billings: Automation

surprises. In: Handbook of Human Factors and Er-

gonomics, 2nd edn., ed. by G. Salvendy (Wiley,

New York 1997)

80.2 J.H.V. Bemmel, M.A. Musen, J.C. Helder: Handbook

of Medical Informatics (Houten, Springer 1997)

80.3 UNECE: United Nations Directories for Electronic

Data Interchange for Administration, Commerce

and Transport (2007)

http://www.unece.org/trade/untdid/welcome.htm

80.4 I. Health Level Seven: HL7 Standards (2007)

http://www.hl7.org

80.5 N.E.M. Association: The DICOM Standard (2007)

http://dicom.nema.org/

80.6 J. Han, M. Kamber: Data Mining: Concepts and

Techniques (Morgan Kaufmann, San Francisco 2001)

80.7 I.H.T.S.D. Organisation: Technical Documents (2007)

http://www.ihtsdo.org/

80.8 B.W. Chaffee, J. Bonasso: Strategies for pharmacy

integration and pharmacy information system in-

terfaces, Part 2: Scope of work and technical

aspects of interfaces, Am. J. Health Syst. Pharm.

61, 506–514 (2004)

80.9 R. Murnane, S. Brown, F. Kerr, M. Meehan, A.

McDonagh, E. Molony, Y. Mulligan: Nursing Work-

load Activity Analysis: A Critical Component of

the Mater Misericordiae University Hospital’s In-

tegrated Nursing Information System (INIS) (2006)

http://www.hisi.ie/html/news.htm

80.10 B.G. Buchanan, E.H. Shortliffe: Rule-Based Expert

Systems: The MYCIN Experiments of the Stanford

Heuristic Programming Project (Addison-Wesley,

Reading 1984)

80.11 S.E. Straus, W.S. Richardson, P. Glasziou,

R.B. Haynes: Evidence-Based Medicine: How to

Practice and Teach EBM, 3rd edn. (Churchill Liv-

ingstone, New York 2005)

80.12 SEER: The Surveillance, Epidemiology, and End Re-

sults (SEER) Program (2007) http://seer.cancer.gov/

80.13 NHIRD: National Health Insurance Research

Database (2003)

http://www.nhri.org.tw/nhird/index.php

80.14 S.A. Ness: Microarray analysis: basic strategies for

successful experiments, Mol. Biotechnol. 36,205–

219 (2007)

80.15 M.W. Isken, B. Rajagopalan: Data mining to

support simulation modeling of patient ﬂow in

hospitals, J. Med. Syst. 26, 179–197 (2002)

80.16 P. Jannin, X. Morandi: Surgical models for

computer-assisted neurosurgery, Neuroimage 37,

783–791 (2007)

80.17 I.H. Witten, E. Frank: Data Mining: Practical Ma-

chine Learning Tools and Techniques, 2nd edn.

(Morgan Kaufmann, Amsterdam 2005)

80.18 E.S. Patterson, R.I. Cook, M.L. Render: Improv-

ing patient safety by identifying side effects from

introducing bar coding in medication administra-

tion, J. Am. Med. Inform. Assoc. 9, 540–53 (2002)

Part H 80

1433

Nanoelectroni

81. Nanoelectronic-Based Detection

for Biology and Medicine

Samir M. Iqbal, Rashid Bashir

This chapter is a review of the work in nanoelec-

tronic detection of biological molecules and its

applications in biology and medicine. About half

of the chapter focuses on the methods employed

to immobilize deoxyribonucleic acid (DNA)onsolid

substrates with particular focus on the electronic

detection and characterization of DNA. Charge-

transfer properties and theories are explained, as

such electronic and electrical sensing of molecular-

level interactions are very important in medical

applications for rapid and cheap diagnosis.

A special tool called nanopore, which has

been used extensively to characterize DNA,is

then reviewed. A special distinction is made

between the characteristics, capabilities, and

impacts of the biological and the solid-state

nanopores. Nanopores, when used in the ion

current measurement setup, are used to measure

the behavior of DNA as it traverses the nanopore.

When the DNA traverses the pore, the blockage of

the ion current is observed as a pulse. The statistical

analysis of the pulses yields trends that are used

to sort the DNA based on various properties. The

nanopores are strong prototypes for biosensors,

and have become a major experimental tool for

investigating biophysical properties of double and

single strands of DNA.TheDNA sequence can

81.1 Historical Background ...........................1433

81.2 Interfacing Biological Molecules.............1434

81.2.1 Guidelines for Preparing Silicon

Chips for Biofunctionalization........1436

81.2.2 Veriﬁcation of Surface Densities

of Functional Layers .....................1437

81.3 Electrical Characterization

of DNA Molecules on Surfaces.................1438

81.3.1 Indirect Measurements

of Charge Transfer Through DNA .....1438

81.3.2 Direct Measurement

of DNA Conduction

and DNA Conductivity Models.........1438

81.4 Nanopore Sensors for Characterization

of Single DNA Molecules ........................1441

81.4.1 Biological Nanopores ....................1441

81.4.2 Solid-State Nanopore ...................1443

81.4.3 The Promise

of Low-Cost DNA Sequencing .........1446

81.5 Conclusions and Outlook .......................1447

References ..................................................1447

potentially be determined by measuring how the

forces on the DNA molecules, and the ion currents

through the nanopore, change as the molecules

pass through the nanopore.

81.1 Historical Background

Nanotechnology, in the last decade or so, has brought

together scientists and engineers from a diverse array

of ﬁelds to collaborate and share their distinct sets

of expertise and tools. The conﬂuence of divergent

technologies has provided better handles on nanoscale

processes and species. Characterization and control at

these fundamental limits have shown immense poten-

tial to improve existing technologies and provide better

tools for understanding at the most basic levels of na-

ture. This has opened horizons wide open for many

diverse areas of research. The major impact it has

had, and will continue to have, is in medical applica-

tions. Biologists have known for decades that cells and

biomolecules are selectively managed, regulated, and

controlled, but the tools to organize and direct these in-

teractions are only now emerging, tools to handle and

Part H 81

1434 Part H Automation in Medical and Healthcare Systems

integrate synthetic structures in diagnostics and thera-

peutics. The nascent tools of nanoscience are helping

unravel the physics of biological processes in unprece-

dented detail. The detection and sensing of biological

entities has enormous promise, and challenges of cor-

respondence size. The challenges are manyfold: the

interface of the wet salty biological molecules to dry

cold solid devices, packaging of the devices so that only

the sensing partis exposed to theanalytes, faithful trans-

lation of the biophysical and biochemical interactions

into electricalor optical signals, selectivity of the device

against the analyte of interestin thepool ofthousands of

entities of no interest, identiﬁcation of the useful signal

amongst unwanted noise, sensitivity of the measure-

ment systemall the way down tothe fundamentallimits,

and repeatability of measurements. For any system to

interrogate the few copies of the analyte of interest re-

producibly and reliably, the interface also plays a key

role. These challenges are the major impediments to the

integration of nanotechnology in biologicaland medical

applications.

A few industries that use a combination of en-

gineering, physics, chemistry, and biology in very

nontraditional approaches to develop devices for med-

ical applications have very recently emerged. Some

examples are on-chip DNA hybridization and opti-

cal detection, DNA detection at the nanoscale for

diagnostics in medical, military, and food safety ap-

plications, etc. DNA carries the genetic information

for living organisms and has a special, important

place in many areas such as genetic engineering, drug

discovery, and gene therapy. The methods used for

immobilizing DNA on solid substrates and the elec-

trical sensing schemes will be detailed in ensuing

sections in this chapter. As we go along, we will

focus on biological nanopores, also known as ion chan-

nels, which are integral parts of the living organisms

at various levels. These have inherent capabilities for

ion-transport regulation under transmembrane poten-

tials. Ion channels have been used to measure the

translocation behavior of DNA as it passes through

the channel. These form an important framework for

DNA detection, but have inherent issues and problems

regarding the stability. Solid-state nanopores, direct

analogues of the biological ion channels, are then

reviewed. Solid-state nanopores are stable under vari-

ous environments. Cheap and rapid DNA sequencing

using solid-state nanopore channels is the focus of

a number of major initiatives, owing to the promise of

focused and effective treatment of diseases, and fast

genetic analysis in areas such as forensics and legal

examination.

81.2 Interfacing Biological Molecules

This section is an overview of the techniques used to

probe a single molecule, i.e., to make a stable and

reproducible interface with a biomolecule. It is impor-

tant to make interface contacts to biological molecules

for reproducible and robust interrogation and charac-

terization. Various techniques and chemistries are used

to attach biological molecules, such as DNA, to solid

substrates, either for conductivity measurements, self-

assembly of devices, for bottom-up device fabrications,

for patterning matter on the nanometerscale, in biotech-

nology, and even as biosensors. In the crafty bottom-up

paradigm, DNA has played avital roledue toits speciﬁc

base-pair interactions.

At around the dawn of this century, the major thrust

in nanotechnology came from the Moletronics program

of the Defense Advanced Research Projects Agency

(DARPA) and the National Nanotechnology Initiative

(NNI) of the US government. The objective of the Mo-

letronics program is to develop functioning prototype

electronic computer processors and memory integrated

on the molecular scale [81.1]. An important avenue

that opened with the vision of Moletronics is towards

the development of nanoscale biosensors that are ro-

bust, small, accurate, cheap, and have high throughput.

A biosensor can be deﬁned as a collection of molecu-

lar recognition sites and the transduction components.

In case of electrical biosensors, the focus has been on

direct measurement of the electrical properties of the

molecular interfacing sites. The size of typical organic

molecules is in the range of 1–10nm. To do any realis-

tic electrical measurements through such small objects,

we need at least two macroscopic metallic electrodes at

the same dimensions.

Self-assembled monolayers (SAMs) of molecu-

les are characterized using a large array of direct

analysis tools, but there are few direct label-free char-

Part H 81.2