Improving quality of life measures by knowledge management methods

1
Improving quality of life measures by knowledge
management methods

• Tamar Kakiashvili1,Waldemar W. Koczkodaj1,3,
  Duncan Matheson1,
• Phyllis Mongomery1, Kalpdrum Passi1, Ryszard
  Tadeusiewicz2
• 1 Laurentian University, Sudbury, Ontario,
  Canada
• AGH University of Science and Technology, Poland
• the corresponding author wkoczkodaj_at_cs.laurentian
  .ca
• Presenter Dr. Tengiz Bochorishvili, Academia
  Podlaska (University of Podlasie), Siedlce,
  Poland

2
Outline

• Medicine and health measures and instruments
• quality of life and WHOQOL
• consistency-driven pairwise comparisons (CDPC)
  can be used to improve WHOQOL measure
• Our approach assumes that not all items in an
  instrument are of equal importance or equally
  contribute to the total score
• it is expected that computing new weights for
  each separate item will contribute to the
  precision of WHOQOL

3
Outline

• Let us look at WHOQOL web site
• http//www.who.int/substance_abuse/research_tools
  /whoqolbref/en/

4
WHOQOL-BREF labels

• Rate QoL
• Q1 - Your quality of life (self-assessment)
• Q2 - Health satisfaction
• Q3 - Pain prevents doing
• Q4 - Need medical treatment
• Q5 - Enjoy life
• Q6 - Life is meaningful
• Q7 - Concentrate
• Q8 - Safe in your daily life
• Q9 - Healthy physical environment
• Q10 - Energy for everyday life
• Q11 - Accept bodily appearance
• Q12 - Money to meet needs

5
WHOQOL-BREF variable labels

• Q13 - Information available
• Q14 - Opportunity for leisure activities
• Q15 - Able to get around
• Q16 - Sleep
• Q17 - Daily living activities
• Q18 - Capacity for work
• Q19 - Satisfied with yourself
• Q20 - Personal relationships
• Q21 - Sex life
• Q22 - Friends
• Q23 - Conditions of your living place

6
WHOQOL

• The quality of life definition proposed by World
  Health Organization (WHO) is important because of
  its universal acceptability.
• It is a combined effort of 15 research centers
  who jointly developed a quality of life
  instrument.
• unfortunately both Canada and Poland are not
  represented by their research teams but hopefully
  not for long

7
The pairwise comparisons preliminaries

• Pairwise comparisons method creates a matrix (A)
  of values (aij) of the i-th object compared with
  its corresponding (j-th) object (in our case,
  questionnaire item)
• PC matrix is reciprocal (aij) 1/(aji) since i
  to j is (or at least, is expected to be) the
  reciprocal of j to i.
• A small scale 1/c,c is used for for comparisons
  from 5 to 9 (in most practical applications)
• inconsistency analysis is essential since it
  allows to locate the most inconsistent input

8
Inconsistency analysis in brief

• The first ever defined in 1977 by Saaty as a
  deviation of an eigenvalue of the pairwise
  comparing matrix.
• Eigenvalue is a global characteristic of a matrix
• It does not localizes inconsistency
• In 1993, a localizing distance based
  inconsistency definition was proposed by
  Koczkodaj

9
What is inconsistency?

• It occurs when we have at least three object to
  compare.
• For example taking into consideration their area,
  we may come to the following input

10
What is inconsistency?

• The previous picture is fuzzy since our knowledge
  is anearly always fuzzy, inexact, incompletem,
  and inaccurate
• by looking at the above picture, one can even
  wrongly conclude that if a is two times bigger
  than b and b is three times bigger than c
  than a should be 6, not 5, times bigger than
  c.
• Why? Our input in undisputable! We really do not
  know if 3 is correct or not (it could be 2.5) or
  2 could be 5/3!

11
What is inconsistency?

• Our comparisons can be subjective and in the case
  of QOL, even highly subjective. Often, we cannot
  be even sure of them. This is a true case of
  approximate reasoning and incomplete
  knowledge
• for a triad, we define inconsistency index as
  follows

and it is the minimum distance from the nearest
consistent triad It is not the purpose of this
presentation to explain why or even how the
above inconsistency index works but how to use it
to improve the QOL measure by using the
inconsistency index
12
Let us begin with a basic model

• Explanation all WHOQOL items have been organized
  into hierarchical structure

13
So let us do it by looking at several screen
images

• This is a part of a model for WHOQOL measure
  since the full model has not fit into the screen

14
screen images

• This is how we compare objects against each other
  starting from the top level

15
screen images

• After a new value (for a pair or more) is
  entered, we close this window and

16
screen images

• We can check inconsistency by ANALYZE window
  selecting Show Max here

17
screen images

• We can alter values of the most inconsistent
  triad (the dark highlight)

18
screen images

• It brings us to this (previously shown) screen

19
screen images

• We can check the inconsistency again

20
screen images

• When we have inconsistency 1/3 or less (but do
  not even try to get not zero since it is
  unrealistic for highly subjective values), we
  obtain weights

21
In conclusion

• Many health measures, including the WHOQOL, do
  not have weights for items
• Applying the pairwise comparisons method to
  WHOQOLhelps to improve its precision as the
  preliminary results show (slide or screen image)
  shows.
• There is very little dispute that pain is more
  important for QOL than transportation

22
In conclusion

• It is a work-in-progress report
• an initial undertaking is the examination of the
  feasibility of CDPC method to improve the
  precision of QOL
• for the final results to be clinically
  applicable, true assessments must be consulted
  with the WHOQOL experts and it may not be easy

23
Professor Koczkodaj

• will answer your questions now