Skip to content

Test 5 - Sex Ratio

Overview Here we compare score distribution to check sex fairness, and similarity to reference  

Parameters (see env.sh) All parameters are included in env.sh and described in External Silent Run. In particular, this test uses:

  • SCORE_MIN_RANGE, SCORE_MAX_RANGE - range of score to inspect, will jump with 0.01 between scores
  • Assumes test 02 completed and we have "$WORK_DIR/compare/rep_propensity_non_norm.matrix","$WORK_DIR/compare/rep_propensity.matrix", "$WORK_DIR/compare/test.preds", "$WORK_DIR/compare/reference.preds" Alon - we have not discussed those in test 2, what are they and are they important to mention at all?

What is actually done? Compare results (reference vs new dataset) by sex and positive rate.   How to read the results? The following is example of results for Test 5 (see 05.sex_ratio.log) For each score cutoff in range, we see:

  • The flagged males ratio in this cohort VS reference. . For instance, we can see here:
    • Male ratio going down when cutoff goes up, both for reference and test.
    • However, the values are higher for the tested dataset, probably because of overall bigger share for males (need to verify).
  • Positivity rate, same as we saw in Test 3.