MRC BSU @ Cambridge Science Festival
The Wonders of Biostatistics
at 2012 Cambridge Science Festival
 |
Saturday 17th & Sunday 18th March at Biology Zone
The MRC Biostatistics Unit presented 'The Wonders of Biostatistics' to Cambridge Science Festival audiences who visited the Biology Zone at the New Museums Site.
'The Wonders of Biostatistics' demonstrsted how key ideas in biostatistics are used to understand, and improve, the public health. Different activities allowed visitors to apply these ideas for themselves in a series of challenges!
The activities covered four different, complementary themes:
1. How many ducks are there?
(Capture-Recapture Studies: used in public health for counting hard-to-reach sub-populations) 
Activity organiser: Simon White
A common question in public health research is how many people have a specific illness/condition or belong to an at risk group. This information is used to inform policy and for monitoring public health interventions.
Since researchers can't find everyone individually, especially for hard to reach groups within the population, one possibility is a capture-recapture study wherein several separate 'captures', such as in reports of assault to police, refuges, and hospitals, are use to estimate the totality of assaults, some of which will have been unreported in all three 'captures'.
Attendees worked in pairs and each captured a sample of ducks in a race against the clock. We then combined these separate captures using statistical methods and worked out how many poorly ducks there were- without having to count all individual ducks.
2. Looking after your hands (Statistical Modelling of Arthritis)
Activity organisers: Brian Tom and Aidan O'Keeffe
How do we move our fingers? What happens if our joints cannot move properly? What tasks would we find difficult to do? Answers to these questions help to explain what arthritis is and how it can affect people's lives.
When treating arthritis, doctors aim to prevent joint damage from happening. Researchers use statistical models, and data collected from arthritis sufferers, to help understand the disease and predict what may happen to patients in the future.
Visitors were invited to try on specially designed gloves that demonstrate how arthritis can affect their hands and impact on daily life. Participants also saw how statistical models are used to predict future damage to the joints of the hands based on the type of arthritis and on the joints already damaged.
3. Piecing together the jigsaw puzzle in "meta-analysis"
Activity organiser: Rebecca Turner
Every year, scientists from across the world carry out hundreds of medical studies. They want to find out which treatments work best for all kinds of illnesses.
Often, there are multiple studies researching the same topic. These sometimes give very different answers!
Scientists need to put together the results from similar studies, like a jigsaw puzzle, so that doctors can see the whole picture. This is called "meta-analysis".
In this activity participants were asked 'Does brushing our teeth really work?'
Researchers carried out a meta-analysis to investigate how much tooth decay is prevented when children brush their teeth every day with fluoride toothpaste. They found 70 studies reporting an estimate of how much tooth decay was prevented when children used fluoride toothpaste instead of fake (placebo) toothpaste.Attendees tried out a giant jigsaw puzzle to find out the answer and learnt how important "meta-analysis" is in helping to improve health.
4. How random are you? (Use of random numbers in statistics)
Activity Leader: Ruth Keogh
In statistical investigations 'random' numbers are widely used, but to use them we need to be able to generate them. How do we generate numbers randomly from a particular set?
A simple example is that a die can be used to generate random numbers from the sequence 1, 2, 3, 4, 5, 6. On any given throw, any of the numbers should be equally likely to appear. If we throw the dice many times, we expect to get each of the numbers 1, 2, 3, 4, 5, 6 to appear equally-often.
In this activity we investigated how good humans and dice are at generating random numbers from the sequence 1 to 6.
Attendees got to play with giant dice, become a human die, and learn about randomness.
BSU at the 2011 Cambridge Science Festival
Members of the MRC Biostatistics and Human Nutrition Research Units will show their joint interactive stand 'Are you really what you eat?' to the Cambridge Science Festival audience visiting the Biology Zone at the New Museums Site on Saturday the 19th of March.
The exhibit based on body composition, will be the same shown in the Big Bang Fair in London, and will have visitors trying foods in a secret challenge and measuring their own personal body composition recipe. They will also have the chance to see how their body recipe change as they get older and how what foods they choose for dinner are related to their health not just today, but in 10years time as well.
The BSU will also run a selection of two interactive exhibits:
- Size matters!
When designing a randomised controlled trial (RCT, or other experiment), a key decision is how many patients (or individuals) need to be recruited
Size matters: so that, despite random variation, if a new drug is better than usual treatment to the extent that is biologically plausible, that superiority will be clear from the RCT. That is: when the yardstick of statistical significance is used to compare outcomes between those who were randomised to receive the new intervention versus conventional treatment.
How is study size worked out? In complex trials, when sums are too hard, we can use random numbers to simulate the probable outcomes on new vs. usual treatment.
The Size matters! stand visitors will take part in a simulation: work out how many offenders should be randomised between drug-rehabilitation-order (DRO) versus prison (for 180 days), to find out if the proportion of offenders with no further convictions in the 2 years after randomisation increases from 30% for those jailed to 40% with DRO.
- Comparing survival in randomised trials
If we are not sure whether a group of patients has better survival after surgery or with medical treatment we can assess it using a randomised controlled trial. Patients enter the trial over time, have their allotted treatment and are followed up for survival.
Survival methods are a set of mathematical tools to deal with this situation.
Important questions are:
- What is the time origin? Time since study started? Time from randomisation? Age?
- How to deal with incomplete follow up? (censoring)
- Is early risk of death from surgery outweighed by longer term reduction in risk?
- How long should we follow up patients before analysis? We need answers as quickly as possible but early stopping may give the wrong answer
In this interactive activity, the visitors will conduct their own randomised controlled trial on the computer, with a programme designed by the BSU team.
For further information about the Cambridge Science Festival click here.