Skip to main content
. 2018 Jan 30;5(1):e000240. doi: 10.1136/bmjresp-2017-000240

Table 3.

Dichotomised survival prediction models were generated using Lasso regression in 269 patients with malignant pleural mesothelioma

Model 2 (survival <6 months) Model 3 (survival <12 months)
Optimal threshold: 0.3 Optimal threshold: 0.6
False True False True
False 40 10 False 27 22
True 19 28 True 7 38

The performance of dichotomised predictions (rows) at selected optimal threshold values relative to the observed survival outcomes (columns) is reported in contingency tables.

Lasso, least absolute shrinkage and selection operator.