Inside the web site will be a database of all known, relevant data and studies and a probability calculator that uses advanced Bayesian stastistics and information-theory based AICc model comparison methods to create the most accurate possible regression model. What sort of data is needed? The data gathered will be relevant to the following questions…
- If person A has X and person B doesn’t, what is the chance Z of person B getting X via activity Y?
- What is the list of sexual activities that can transmit different diseases and what are the different variations in the behavior or forms of protection while performing the act that may affect transmission?rough vs gentle, forms of barrier, etc
- How does number of times performing act Y affect the per act chance of transmission, Z? In one extreme, simply performing the act once exposes B to the complete transmission risk, and further performances of the act add no further risk. In the other extreme, each performance of the act is an independent chance of transmission.
- What is the variance within the population of susceptibility? Of transmissibility? This variance could be important for estimating transmission possibilities (ie average transmission rate is not adequate).
- What are known traits of persons A and B, physical or behavioral, that affect transmission rates?
- What is the infection rate of X in the population?
- How does infection rate of X vary across different demographic variables and by region? What are the important demographic variables to consider? What are the demographic variables that have been researched?
- Are there any transmission relevant behavioral or physical traits that vary across demographic lines?
- What is the clearance probability over time for the STD?
There will be a team of dedicated researchers collecting and organizing the data base and updating the probability calculator. In tandem, the idea will be explored to have a moderated research-community wiki for updating the database: filling in gaps, updating estimates, and identifying dimensions that should be added to aid in prediction. Having such an open-source approach has a lot of benefits and some known problems, so in the short run, this would have to be seen as a way to help the dedicated team while the idea of having a moderated version of the wiki flowing into the data base is explored.
There is a lot of information out there that is not easily accessible. The web site will be a clearing house for this information and the citations of original studies… a kind of one-stop-shopping-center for STD research. There are also a really important set of gaping holes in research. The construction of this probability calculator will expose these holes. There are places where there is no truly relevant data and the best that can be offered are priors based on back-of-the-envelope or even out-of-thin-air estimates. Where necessary to create predictions, they will be derived in as principled a way as possible. All parameters will have easy to find links to their derivations, including citations of relevant original research and metastudies and straightforward statements about where ‘guesstimates’ were used and the impacts of these guesstimates on prediction dynamics. The data, guesstimates, and the impacts of guesstimates will all be easily found through the Back End.