microbiology and computer

[penguin]

I did a science project last year. I looked at whether routine testing of methicillin resistant staphyloccus aureus can be cost effective. I made a computer model using a commerically available software model and ran simulations. I did not medal . Unfortunately, I did not get any judge's feedback. I am still working on a similar topic. Any thoughts on what I can do to improve?

[MelissaB]

Hi,

Unfortunately, without seeing your board, it's hard to tell what you could do to improve. It sounds like a great project, so I think it may have been a presentation issue--alternatively, it may be because you did not do what they consider an 'experiment', with independent and dependent variables. (I personally think it sounds like a great project!).

What are you planning to do for this year? Once we know that, we can better help you.

[penguin]

Let me tell you what I did last year since this year's project is still in research phase. It will essentially be similar to last year's project. So let me try to explain my last year's project.

There have been increased reports of Methicillin-resistant Staphylococcus aureus (MRSA) infection especially in high school athletes involved in contact sports. My project examines whether testing all skin infections in this population is worthwhile. A realistic life-like computer decision model was constructed. Then, simulations were performed using currently available data found in the literature search, including prevalence rates, therapeutic options, possible outcomes, and costs. It was found that at the current prevalence of MRSA, testing is not cost-beneficial. However, as the prevalence increases, testing becomes more cost-effective. My data analysis was quite complicated. It measured the amount of money it costed per number of QALY (quality adjusted life years) it saved to implement the intervention, therefore cost-effectiveness. Does that make sense? This year, instead of measuring QALY, I will actually measure the number of infections that are prevented.

[aelin]

Hi,

That does sound like a very important question, but keep in mind that getting a medal is not an indication of how "good" a research project is. Nevertheless, since medals are (sometimes) shiny, there are always ways to improve any project.

I don't see any a priori problems with the topic or the way the research was conducted, as it is definitely an experiment, contra to what was potentially suggested before. However, as I myself did a computer simulation for my project a while back, I will offer my few cents worth. There is a famous saying about models, something along the lines of "models are never accurate, but sometimes useful." I think you really want to stress this sort of thing during your presentation, as I ran into trouble a bit with this as well. It is perfectly fine that you ran your simulations with current clinical data (this is in fact exactly what you want to do), but you also need to show that you model is valid. That is, prove to the judges/audience that your model makes predictions that are in accord with what appears to be occurring in the real world. So, if possible, you also need to run simulations on past data (say 2005 or so), and show that the predicted results for 2007 (per your model) match up (at least to some degree) with the actual results for 2007, and etc. In this case, the numbers don't have to be perfect; you just have to show that your model is useful.

As a small side note, my gut instinct is that measuring cost-effectiveness with QALY instead of infections prevented would make more sense since cost-effectiveness is more of an economic choice. However, as I don't fully understand how the model works, it is up to you to decide how to implement it, so long as it makes sense to you and you can explain it.

Hope this helps!
Aaron Lin