Research in Health Services

1
Research in Health Services

• Concept and Idea
• Rajnish K Gupta
• Professor of Cancer Studies

Slides Aisling de Hooge Research and Information
officer Cancer Services
2

• Search and Literature
• Hypothesis Rationale

3
Search and Literature Review

• In information retrieval systems, the match
  between query and document is severely unbalanced
  due to the huge difference in the sizes of query
  and document.
• Strengthen the query with related knowledge such
  as context semantics
• Context based search approach
• Conceptual paradigm for performing search in
  context
• e.g. IntelliZap system

4
Search and Literature Review

• Knowing where to look
• An overview of search engines

5
General Search Engines

• Google
• http//www.google.com
• Yahoo
• http//www.yahoo.com
• Askjeeves
• http//www.askjeeves.com
• All the Web
• http//www.alltheweb.com
• AOL
• http//search.aol.com/
• Hot Bot
• http//www.hotbot.com

• Teoma
• http//www.teoma.com
• Alta vista
• http//www.altavista.com
• Gigablast
• http//www.gigablast.com
• Lycos
• http//www.lycos.com
• MSN
• http//search.msn.com

Medical Search Engines
6
Google http//www.google.com
7
Yahoohttp//www.yahoo.com
8
Askjeeves http//www.askjeeves.com
9
All the Webhttp//www.alltheweb.com
10
AOLhttp//search.aol.com/
11
Hot Bothttp//www.hotbot.com
12
Teomahttp//www.teoma.com
13
Alta vistahttp//www.altavista.com
14
Gigablasthttp//www.gigablast.com
15
Lycoshttp//www.lycos.com
16
MSNhttp//search.msn.com
17
Medical Search Engines

• Health On The Net MedHunthttp//www.hon.ch/MedHu
  nt/
• MedHunt uses both humans and web crawling to
  build its index of medical information.
• MedicineNet.comhttp//www.medicinenet.com/script/
  main/hp.asp
• Medical information contributed by over 50
  doctors and health professionals.
• MedlinePlushttp//medlineplus.gov/
• Medical information from the US National Library
  of Medicine and the US National Institutes of
  Health.
• OmniMedicalSearch.comhttp//www.omnimedicalsearch
  .com/
• Meta search major medical search engines and
  databases from this new service.
• WebMDhttp//www.webmd.com/
• Long-standing portal of health and medical
  information.

Science Search Engines
18
Health On The Net MedHunthttp//www.hon.ch/MedHu
nt/
19
MedicineNet.comhttp//www.medicinenet.com/script/
main/hp.asp
20
MedlinePlushttp//medlineplus.gov/
21
OmniMedicalSearch.comhttp//www.omnimedicalsearch
.com/
22
WebMDhttp//www.webmd.com/
23
Science Search Engines

• Biocrawler http//www.biologie.de/
• Directory and search engine for biological
  information.
• Chemie.DE
• http//www.chemie.de/
• Directory and search engine for information
  about chemistry.
• Scirus http//www.scirus.com
• Scirus combines a targeted crawler from FAST
  that focuses only on web sites with scientific
  content.
• Search4Science http//www.search4science.com/
• Search4Science is a search engine put together
  by scientists for scientists. It shows -- if you
  enter simple keyword searches you're often
  presented with related scientific terms to expand
  or limit your query.
• SciSeek http//www.sciseek.com/
• SciSeek is a focused web directory created by
  human editors. It's a useful tool for browsing
  for information in a specific scientific area.

Other websites
24
Biocrawler http//www.biologie.de/
25
Chemie.DEhttp//www.chemie.de/
26
Scirus http//www.scirus.com
27
Search4Science http//www.search4science.com/
28
SciSeek http//www.sciseek.com
29
Other Websites

• PubMed
• http//www.ncbi.nlm.nih.gov/entrez/query.fcgi
• World Health Organisation
• http//www.who.int/en/
• National Library of Medicine
• http//www.nlm.nih.gov/nichsr/hsrsites.html
• Health Finder
• http//www.healthfinder.gov/
• Irish Health Research Board
• http//www.hrb.ie/

• Health Web
• http//healthweb.org/
• Discovery
• http//health.discovery.com/
• Irish Health
• http//www.irishhealth.com/
• Health World Online
• http//www.healthy.net/
• VHI
• http//www.vhihealthe.com/
• Bupa
• http//www.bupa.ie/
• Irish Medical Organisation
• http//www.imo.ie/

Hypothesis Rationale
30
Pub Medhttp//www.ncbi.nlm.nih.gov/entrez/query.f
cgi
31
World Health Organisationhttp//www.who.int/en/
32
National Library of Medicinehttp//www.nlm.nih.go
v/nichsr/hsrsites.html
33
Health Finderhttp//www.healthfinder.gov/
34
Irish Health Research Boardhttp//www.hrb.ie/
35
Health Webhttp//healthweb.org/
36
Discoveryhttp//health.discovery.com/
37
Irish Healthhttp//www.irishhealth.com/
38
Health World Onlinehttp//www.healthy.net/
39
VHIhttp//www.vhihealthe.com/
40
BUPAhttp//www.bupa.ie/
41
Irish Medical Organisationhttp//www.imo.ie/
42
Hypothesis Rationale
43
Plato

• Several levels of comprehension ranging from
  those based purely on sensory impressions to
  eventual understanding, through dialectic, of the
  first principles or Eidos (Ideas, or Forms)

44
Plato
45
Whewell

• Natural Philosopher Scientist
• Imprecise nature of Ideas, Forms.
• Modern physics can be tested by the emergence of
  scientific methodology.
• Thought alone can never approach reality unless
  founded on fact.
• Such facts, referred to as data (datum) are
  derived from the observation of natural events
  or, from carefully designed experiments and
  executed interventions.

46
Fibonacci

• Liber Abaci described the famous Fibonacci
  series
• The Divided Line simile superimposed

47
Data versus Doxa

• Plato - in his paradigm of the line, separated
  knowledge (Episteme) from opinion or supposition
  (Doxa)
• Doxa derives purely from sensory impressions
  i.e. from the appearance of things
• Pistis direct experience results in belief
• Eikasia impressions or illusions
• Platos message knowledge cannot be based on
  unsubstantiated belief or conjecture.

48
Science is driven by hypothesis

• A proposed explanation for a set of observations
  (data)
• The ability to use reason to create hypotheses
  goes beyond mathematics
• Perhaps corresponds to Platos dialectic
  ability to reason

49
Science is driven by hypothesis

• It is the testing of such hypotheses by seeking
  additional data (re-search), experimentally
  (where necessary), that distinguishes scientific
  knowledge from beliefs that are not based on the
  rational interpretation of factual information

50
Medical Treatment

• In the past based on Doxa unsupported by data
  e.g. purging, bleeding, trephining
• Successful prevention or treatment of disease
  does not necessarily require detailed
  understanding of the cause, or of therapeutic
  mechanisms.
• Empirical observations provide raw material for
  the development and testing of hypotheses

51
Medical Treatment

• Validity of a hypothesis is examined by making a
  judgement as to whether the relevant data support
  or refute it.
• Occasionally, the result of an experiment or
  trial, designed to test a hypothesis is
  sufficiently obvious no further analysis
  required
• Since different informal observers may draw
  different conclusions from the data set,
  Scientific method - use of objective methods to
  measure the degree of certainty, i.e., the
  probability, that a hypothesis is correct is
  often required

52
The Art of Conjecturing

• Probability theory has become a critically
  important branch of mathematics and is as
  important to the design of clinical experiments
  involving human subjects (clinical trials) as it
  is to the analysis of the results obtained
• Foundations were laid by Blaise Pascal and Pierre
  de Fermat in 1654, dealing with the mathematics
  of games of chance e.g. Double six
• Jakob Bernoulli - Bernoulli trial, dichotomous
  outcome e.g. tossing a coin

53
Clinical Trials

• Designed to test the value of a particular
  treatment
• Outcome can be dichotomous e.g. survival or death
  
• Usually more complex since there are many factors
  (or variables) which determine response
• Various characteristics
• Dependent
• Independent
• Mathematical approach multivariate analyses
• Independent risk factors
• Predict prognosis, particular disease, with a
  particular treatment
• Such information permits treatment decisions to
  be made i.e. evidence based

54
Clinical Trials

• Probability theory calculate the number of
  patients required
• Power of a study degree of certainty that the
  difference between the interventions, if present,
  would be detected, specifying the anticipated
  difference and the degree of certainty required
• Bernoullis Law of Large Numbers states that the
  more trials that take place, the closer the
  proportion of successes will be to the proportion
  that applies to an individual trial
• Quantification of risk
• Evolved from Pafnuty Chebyshev - observed value
  is compared to the expected value ie. Relative
  risk

55
So, finally

• There is a proposal to offer coverage for a
  breast cancer screening program to women aged
  20-40 in your Health Board.
• Four statements from four randomized controlled
  trials
• On the basis of each statement, you should
  indicate if you agree to the implementation of a
  breast screening program.
• Assume that the costs of each program are the
  same, each result was deemed to be statistically
  significant.

56
During a 7-year follow-up

• Program A reduced the death rate from breast
  cancer by 33
• Program B produced an absolute reduction in
  deaths from breast cancer of 0.06
• Program C increased the rate of survival from
  breast cancer from 99.82 to 99.88
• Program D prevented one death from breast cancer
  for every 1666 women screened
• Which program do you recommend?

57
Thank you