Evolution and emergence of novel human infections

Arinaminpathy, N.; McLean, A. R.

Emerging human infections N. Arinaminpathy & A. R. McLean 3941 algorithm performance, calculated over 250 emergences simulated as described above, under both the gradual and punctuated scenarios. Both single and double approaches show a similar performance under the punctuated scenario. However, under the gradual scenario, the double approach shows an appreciably better performance in sensitivity, also doing so more consistently. It raises the alarm after a mean (standard deviation) of 4.2 (1.3) cases, compared with 5.8 (2.4) for the single approach. As might be expected, this comes at the expense of some specificity: the single approach causes a mean (standard deviation) of 0.68 (0.97) false alarms before emergence, compared with 1.0 (1.3) for the double method. In practice, epidemiological investigation and contact tracing will be essential for identifying which cases are epidemiologically linked. Furthermore, we do not address here the timing or distribution of index cases. Nonetheless, while more sophisticated schemes are undoubtedly possible, these results illustrate how additional outbreak characteristics, such as incidence, could contribute to emergence detection.