This project is the basis of your grade for the course. You are
encouraged to get repeated input from the instructor through out the semester.
The project should provide a detailed probabilistic risk analysis for an actual
organization or situation. Here are some examples:
- Analysis of privacy risks within an organization (e.g. release of
information through employee sales)
- Analysis of security violation risks (e.g. school shootings)
- Analysis of risk of sentinel events (e.g. wrong side surgery, epidemic
infections, patient elopement, etc.)
- Analysis of causes of an outbreak of a new disease
In case you find it difficult to arrange for access to information within an
organization, you might want to focus on risk of relapse to poor habits within a
personal improvement effort:
- Analysis of causes of failure of women with cardiovascular illness for
maintaining recommended exercise and diet
- Modeling a patient's potential reaction to changes in medication dose or
The project must include a mathematical model of
risk. It must provide data validating assumptions made. Data could
come from experts in the field, the literature and/or from surveys.
A midway peer review is required. Prepare
a report that includes all sections required in the final report. Send
your work to another student and ask for comments. Reviewers must respond
to both the student and the instructor and must include in their review comments
on the following:
- Does the title describe the content?
- Is the abstract complete?
- Is the writing clear, grammatically
correct and free from spelling errors?
- Is the visual presentation of the model
understandable by itself without reference to the text?
- Has the report followed the recommended
- Is there a literature review of the
- Is the mathematical mode correctly
- Is it clear how model structure was
- Is it clear how parameters of the
model were estimated?
- If experts were queried, was
the question asked in a way that limited the population of
concern to the patients or situations with appropriate
- If time between events was
used, was the average time between consecutive events correctly
- If data were constructed from a
series of failures and their causes, was the joint distribution
of causes and sentinel event correctly calculated.
- Has the validation of the model
parameters and structure been carried out, including:
- Conditional independencies have
- Predicted frequency of sentinel
event is within scale of observed frequency
- Prevalence of causes among sentinel
events is as experienced.
- What worked well in the report that you
can use in your own effort?
- What needs improvement that could make
the report more useful.
Peer reviews should follow the outline provided
above exactly and should have a section for each topic area, including the
number and the relevant alphabet. There should be a total of 10 labeled
sections in the peer review.
The following components should be included in
the final report:
- Title page (1 page)
- Title of the report
- Author names (with highest academic
degrees) and affiliations (including titles, departments, and name and
location of institutions of primary employment)
- Corresponding authorís name and
complete address including email
- Any acknowledgments credits, or
- Abstract (200 words or fewer)
- What was investigated
- Methods of analysis
- Sources of data
- Results of validating the model
- 3-5 key words that describe the
contents of the article
- Introduction (1 page)
- Why are these risks of concern
- How would the risk analysis help in
- Why would a probabilistic risk analysis
be preferred in this setting
- Methods (2-3 pages)
- Procedures used to construct the risk
- Procedures used to estimate the
parameters of the risk model
- Procedures used to validate the
- Sources of data (1 page)
- Indicate source of data for each
parameter including experts opinions, literature review, extant
databases, brief surveys, diary.
- Results (2-3 pages)
- Describe the model
- Describe the findings from test of
validation of the model
- Show predicted event is within
observed frequency of the event
- Show prevalence of causes are
within reported frequency of these causes in the literature
- Show result of tests to confirm
accuracy of assumed conditional independencies
- Conclusions (1/2 page)
- Discus the utility of the model
- Discuss areas of additional work
Farrokh Alemi, Ph.D. Created on Tuesday October
4th, 2006. Most recent revision
10/22/2011. This page is part of a
Course on Risk Analysis