Skip to main content
. Author manuscript; available in PMC: 2020 Oct 1.
Published in final edited form as: Lancet HIV. 2019 Jul 5;6(10):e696–e704. doi: 10.1016/S2352-3018(19)30139-0

Table 4:

Performance of LASSO algorithm for detecting incident HIV infection in the development cohort.

Percentile of HIV risk score used to
define test positivity		 Risk score (out of
100,000)		 Sensitivity
(%)		 Specificity
(%)		 Positive predictive
value (%)		 Negative predictive
value (%)a
10% 0 100 65·4 0·04 100
20% 0 100 65·4 0·04 100
30% 0 100 65·4 0·04 100
40% 0 100 65·4 0·04 100
50% 0 100 65·4 0·04 100
60% 0 100 65·4 0·04 100
70% 1 96·0 70·3 0·04 100
80% 2 94·7 80·4 0·06 100
90% 8 77·3 90·0 0·10 100
91% 9 71·3 91·0 0·10 100
92% 10 70·0 92·0 0·11 100
93% 11 70·0 93·1 0·13 100
94% 12 68·7 94·2 0·15 100
95% 13 67·3 95·0 0·18 100
96% 15 64·7 96·3 0·22 100
97% 18 56·7 97·0 0·25 100
98% 25 50·7 98·0 0·33 100
99% 32 41·3 99·0 0·54 100
a

High negative predictive values reflect the low HIV incidence in the population.