Analytical validation of Exandra: a clinical decision support system for promoting guideline-directed therapy of type-2 diabetes in primary care – a collaborative study with experts from Diabetes Canada

Table 5 Calculated Exandra precision

Response to PP case	Total (N = 946)	Chapter 1. Glycemic management (n = 293)	Chapter 2. Dyslipidemia (n = 297)	Chapter 3. Blood pressure control (n = 297)	Chapter 4. Anti-platelet therapy (n = 109)
Yes (TP)^a	894	263	276	267	88
No (FP)^b	42	15	5	9	13
Not sure^c	10	0	1	6	3
Exandra precision ratio (%)^d	95.5	94.6	98.2	96.7	87.1

FP false positive, PP predicted positive, TP true positive
^aYes (TP) = 894 because 556 “yes” from Phase 1 (Table 3) + 228 “no” cases changed to “yes” from Phase 2 (Table 4) + 110 “not sure” cases changed to “yes” from Phase 2 (Table 4) = 894
^bNo (FP) = 42 because 26 unchanged “no” cases from Phase 2 (Table 4) + 16 “not sure” cases changed to “no” in Phase 2 (Table 4) = 42
^cNot sure = 10 because 6 unchanged “not sure” cases from Phase 2 (Table 4) + 4 “no” cases changed to “not sure” in Phase 2 (Table 4) = 10
^dPrecision = TP/PP = TP/(TP + FP). This calculation excludes the “not sure” cases

ISSN: 1472-6947