Estimating Trends in HIV Incidence for Australian Women
The work is concerned with a project funded by an NHMRC grant for which Prof Niels Becker, NCEPH, is the principal investigator and Dr. Zhengfeng Li is a team member. The aim of the project is to apply a newly developed method, using linked data on HIV and AIDS diagnoses, to reconstruct the infection curve and thereby estimate trends in HIV incidence for Australian women who were infected heterosexually. In mathematical terms, we are estimating curves and computing the precision of the estimated curves. The computations use an algorithm adapted from the EM-algorithm, by adding a smoothing step and a Newton-Raphson step.
|
Principal Investigator Niels Becker |
Project x00 Facilities Used SC |
|
Co-Investigator Zhengfeng LiNCEPH ANU
|
RFCD Codes 239901 |
Significant Achievements, Anticipated Outcomes and Future Work
We find that estimating the extra parameters is feasible, producing estimates with good precision. Including age as a covariate gives 90 % confidence intervals for the HIV incidence curve that are about 20 % narrower than those obtained when age data are not used. Our finding has been reported in a paper of "Age-specific backproject of HIV diagnosis data" which was submitted to the journal of Statistics In Medicine for publication.
Computational Techniques Used
The code was implemented in C++.
Publications, Awards and External Funding
The project was funded by NHMRC.
N.G. Becker, J.C. Lewis, Z.F. Li and A. McDonald, Age-specific backproject of HIV diagnosis data, submitted to Statistics in Medicine.