Transcript PowerPoint 簡報

Database System
(DBS)
Information Technology
(IT)
Medical Record
(MR)
Electronic Medical Record
(EMR)
Computer Technology
(CT)
(Web; Digital information)
Medical Informatics
(MI)
Data Mining
(DM)
Evidence Based Medicine
(EBM)
Bio-informatics
Knowledge Discovery
Medical Informatics related disciplines
Medical Information System
Time
2000s
Automated
Medical
Record
Computerized
Medical
Record
Electronic
Medical
Record
Electronic
Patient
Record
Electronic
Health
Record
Paper
base
Scanning
the paper
document
Computer
format
Wider scope
of personal
information
Wellness
information
Knowledge System
Admission
Discharge
Clinic
Department
Reporting
Financial mgmt.
Decision making
Data harmonization
Data warehousing
Data mining
Netwoking
World wide data
Public health
Research system
User Driver
Technology Driver
...............
Knowledge Discovery
& Data Mining
Four basic elements of a database system
User: end user; casual user; practitioners
Data: text; graph; image; sound; vedio
Software: general; application
Hardware: CPU; I/O; network
Types of current database system
Hierarchical database
Network database
Object-oriented database
Relational database (DBASE III plus; Clipper; Foxpro; Access)
Advantages of integrated database
Data sharing
Minimal data redundancy
Data consistency
Data standard improvement
Data integrity improvement
Data security
Faster development of new application
Normalization (best table design)
Separate the table until every fields
depend on the primary key only
Primary key (PK)
The group of attributes that
uniquely
(no null, ) determines every other
attribute in the relation Candidate
key
All candidate PK’s are candidate
keys
Foreign key
A attribute value of reference table
is
the PK of the home table
First normal form (1NF)
If and only if 1)all domains contain single
values
only
Second normal form (2NF)
If and only if 1) it is in 1NF, and 2) every nonkey attributes in fully dependent on the PK
Third normal form (3NF)
If and only if 1)it is in 2NF, and 2) every nonkey
attributes is non-transitively dependent on
the
PK
Further normalization
Boyce-Codd normal form (BCNF)
Forth normal form (4NF)
Fifth normal form (5NF)
掛號
健診
藥局
門診
急診
次專科
檢驗
X光
病理
心導管
出院
內視鏡
細胞
血庫
物理治療
營養
病歷
掛號
健診
藥局
門診
急診
次專科
檢驗
X光
病理
出院
心導管
內視鏡
細胞
血庫
病歷
物理治療
營養
供繕
麻醉
住院
手術
護理照顧
掛號
醫師績效
健診
薪資
藥局
門診
急診
次專科
檢驗
人事管理
X光
病理
藥品庫存採購
出院
心導管
內視鏡
財物料採購
細胞
血庫
出納
病歷
物理治療
營養
財務會計
成本會計
供繕
麻醉
住院
手術
護理照顧
醫院資訊系統
1 門急診資訊系統 掛號系統; 醫令作業系統; 收費管理系統; 帳務管理系統; 健保申報系統
2 住院資訊系統
住院管理系統; 批價收費系統; 健保申報申復爭議系統; 單一劑量作業
護理站管理系統; 手術室管理系統; 住院醫囑作業系統; 帳務管理系統;
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
病歷管理系統
藥局連線系統
醫師績效作業系統
供繕系統
血庫作業系統
檢驗檢查報告系統
X 光片管理系統
急診病患動態管理系統
疾病分類統計作業系統
藥品庫存採購作業作業系統
材物料庫存採購作業作業系統
人事管理系統
薪資管理系統
財務會計作業系統
成本會計作業系統
出納作業系統
財產作業系統
健檢系統
掛號
醫師績效
健診
薪資
藥局
門診
急診
次專科
檢驗
人事管理
X光
病理
藥品庫存採購
出院
心導管
內視鏡
財物料採購
細胞
血庫
出納
病歷
物理治療
營養
財務會計
成本會計
供繕
麻醉
住院
手術
護理照顧
Knowledge discovery in databases (KDD)
An explosive growth in capabilities to both generate and collect data
scientific data (e.g. remote sensors, space satellites)
business data (e.g. bar codes for commercial products, credit card)
human genome database (human genetic code)
government transactions (e.g. tax returns)
health care transactions (control costs, improve quality)
advanced data storage technology (faster, higher capacity, cheaper)
better database management systems and data warehousing technology
Allow us to transform this data deluge into “ mountains “ of stored data
Such volumes of data overwhelm traditional manual methods of data analysis
such as spreadsheets, ad-hoc queries
can create informative reports from data
can not analyze the contents of reports to focus on important knowledge
A significant need exists for a new generation of techniques and tools
with the ability to intelligently and automatically
assist humans in analyzing the mountains of data
for nuggets of useful knowledge
These techniques and tools are the subject of the emerging field of KDD
Knowledge Discovery
and
Data Mining
Components
Model representation
Model evaluation
Parameter search & model search
Methods
Decision trees and rules
Neural networks
Regression (linear and non-linear)
Classification and Clustering
Probabilistic graphical dependency
models (Baysian networks)
Relational Learning models
(autoregressive models)
etc.
Applications
Business management
Health care fraud detection & prevention
Astronomy
Molecular biology
Global climate change modeling
Medicine
etc.
Knowledge Discovery
Data preparation
Search for pattern
Data Mining
Knowledge evaluation
Knowledge interpretation
Definitions of KDD terms
Data
is a set of facts (F; e.g. cases in a database)
example: F is the collection of 23 cases with three fields each containing the values of debt,
income, loan.
Pattern
is an expression E in a language describing facts in subset FE of F. E is called a pattern.
example: “ If income <$t, then person has defaulted on the loan “.
Knowledge
A pattern is called knowledge if for some user specified threshold. It is purely user-oriented, and
determined by whatever functions and thresholds the user chooses. By appropriate settings of
thresholds, one can emphasize accurate predictors or useful patterns over others.
Data mining
is a step in the KDD process consisting of particular data mining algorithms that, under some
acceptable computational efficiency limitations, produces a particular enumeration of patterns
over F.
KDD Process
involves data preparation, search for patterns, knowledge evaluation, and refinement involving
iteration after modification.
is the process of using data mining methods ( algorithms) to extract (identify) what is deemed
knowledge according to the specifications of measures and thresholds, using the database F.
is interactive and iterative, involving numerous steps with many decisions being made by the user.
Knowledge discovery in databases vs. data mining (KDD vs. DM)
KDD has been mostly used by artificial intelligence, machine learning researchers
the overall process of discovering useful knowledge from data
DM has been commonly used by statisticians, data analysts, MIS community
application of algorithms for extracting patterns from data
without the additional steps of the KDD process (such as incorporating
appropriate prior knowledge and proper interpretation of the results)
blind application of DM methods can be a dangerous activity
Links between KDD (and DM) and related field
machine learning
pattern recognition
databases
statistics
artificial intelligence
knowledge acquisition for expert systems
data visualization
The National Health Insurance
Information System
Nine major systems:
Underwriting of insurance
Payments for medical fees
Management of medical affairs
Financial management
Administrative support
Decision support system
Exchange of information
Management of community-based insuring agencies
Safety control
健保IC卡存放的資料
健保IC卡資料存放共分四區段，未來視實際需要增加存放內容。
一、「基本資料段」：
主要功能係辨識身分用，存放內容包括：卡片號碼、姓名、身分證字號或
身分證明文件號碼、出生日期、性別、發卡日期、照片、卡片註銷註記。
二、「健保資料段」：
主要功能係紀錄就醫相關資料，包括：保險對象身分註記、卡片有效期限
、重大傷病註記、就醫可用次數、最近一次就醫序號、新生兒依附註記、
就醫資料登錄、就醫累計資料、醫療費用總累計、個人保險費、保健服務
、最後月經開始日期、預產期、孕婦產前檢查。本段資料分階段上線。
三、「醫療專區」：
主要存放門診處方箋、長期處方箋、重要醫令項目、過敏藥物等。本段資
料將考慮院所適應情形，分階段上線。
四、「衛生行政專區」：
主要存放預防接種資料、器官捐贈資料等。
Medical Informatics as a Discipline
(http://www.cpmc.columbia.edu/edu/textbook)
medical informatics =
study and use of computers and information in health care
purpose of this lecture is to further define the field
definition by Asso. of American Medical Colleges (AAMC) 1986
"Medical informatics is a developing body of knowledge and a set
of techniques concerning the organizational management of
information in support of medical research, education, and patient
care.... Medical informatics combines medical science with several
technologies and disciplines in the information and computer
sciences and provides methodologies by which these can contribute
to better use of the medical knowledge base and ultimately to better
medical care."
history of computers
1800s - Charles Babbage's logic engine
1890 - Herman Holleriths's punch cards for census
1940s - early programmable digital computers (Eniac)
1950s - commercially available (Univac)
1960s - faster, more memory
1970s - minicomputers
1980s - microcomputers, networks
1990s - RISC, workstations, growth of networks
appearance of computers in medicine
1960s - practical = early departmental and monolithic
research = early ECG and diagnosis
1970s - practical = monolithic administrative & departmental,
imaging (CT), early bibliographic retrieval
research = alerts, Mycin (early successes)
1980s - practical = results reporting, outpatient,
growth of clinical systems and databases
research = AI, IR
1990s - practical = integration, communication
research = vocab, interfaces, coding, evaluation
factors in lack of use of computers in clinical care
involves complex organisms (unlike physical processes)
if over-simplify, not useful (vs bank transaction)
therefore need sophisticated abstraction + detail
technology for gathering complex info. just emerging
eg low use of QMR or DXplain
therefore providers have not entered info.
reimbursement has not been linked to clinical info.
therefore many admin. systems but few clinical
health care administered by individuals, small groups
less need for coordination
inertia
fear
ignorance
money
security, integrity
lack of standards
language
previous failures
rapid turnover of technology
factors in recent increase of medical informatics
increase in use of technology - more data generated
mobility of population - need to communicate
specialization - need to communicate
managed care systems - need to communicate
rise in health care costs - attempt to control care
improved hardware - faster and more memory
improved methods - acquisition, transfer, retrieval
reduced computer costs
increased awareness
related fields:
makeup of med- info groups
biomedical engineering - ECG, devices
MI higher level of abstraction
electrical engineering - hardware
computer science - algorithms, closer to
mathematics MI specific to health domain
medical computer science –
subdivision of comp sci
cognitive science - AI and psychology
not concerned with studying human brain
information theory - physics of communication
information (library) science –
manage aper/elec info
MI is close to this but MI not limited to
info. storage and retrieval
software industry - producing products
MI stresses evaulation
MI not dependent on selling every roduct
MDs, RNs, dentists,
other health care workers
PhDs, esp computer science (also
physics, ...) administrators,
policy planners
masters, PhD programs in
medical informatics
industry
current issues in clinical care
cost
accessibility of health care
coordinating care and setting policy
acquisition and retrieval of data (eg across inst.)
acquisition and sharing of knowledge (eg specialist)
medical informatics research mirrors clinical issues
data acquisition - GUI, nlp
data storage - databases, modeling
vocabularies - format, content
organization of data - Larry Weed POMR 1969
machine interfaces - standards like HL7, security
data retrieval - query languages
knowledge acquisition
knowledge representation - Arden
application of knowledge when needed decision analysis, alerts, diagnosis
education
care plans and practice guidelines