| Title |
Optimal group testing when prevalence is varying / |
| Translation of Title |
Optimalus grupinis testavimas, kai ligos dažnis kinta. |
| Authors |
Meškonytė, Lina |
| Full Text |
|
| Pages |
66 |
| Keywords [eng] |
Group testing, unknown prevalence, optimisation problem. Grupinis testavimas, nežinomas užsikrėtimo dažnis, optimizavimo uždavinys. |
| Abstract [eng] |
Group (pooled) testing is an alternative to individual (one-by-one) testing that can be used to efficiently (faster and cheaper) test a large set of objects (specimens/samples/individuals/etc.) when samples are pooled in right size groups. The common approach widely investigated in the literature requires to know the exact prevalence rate. However, it is not clear what to do when it is not known. We aim to answer this question by numerically computing the group size which minimises the expected number of tests when there is no prior knowledge of prevalence (except assuming the majority of large tested population is negative/not infected). The optimisation problem is solved using minimax and average squared error loss functions - the optimal group size is calculated for a group of existing probabilistic pooled testing procedures. The research does not cover and could be extended to higher dimensional methods as well as combinatorial procedures, furthermore, not only the average (expected number of tests), but also variance could be taken into account. |
| Dissertation Institution |
Vilniaus universitetas. |
| Type |
Master thesis |
| Language |
English |
| Publication date |
2022 |
