# Biomarker evaluation - c-statistic (AUC) and alternatives

**URL:** <https://discourse.datamethods.org/t/biomarker-evaluation-c-statistic-auc-and-alternatives/6956>\
**Category:** data analysis\
**Created:** [July 25, 2023, 11:30am UTC](https://discourse.datamethods.org/t/biomarker-evaluation-c-statistic-auc-and-alternatives/6956 "2023-07-25T11:30:22Z")\
**Posts on this page:** 5\
**Page:** 3

<div class="post-metadata">

**Author:** ![f2harrell](https://discourse.datamethods.org/user_avatar/discourse.datamethods.org/f2harrell/32/165_2.png) [@f2harrell](https://discourse.datamethods.org/u/f2harrell)\
**Post date:** [March 26, 2026, 12:17pm UTC](https://discourse.datamethods.org/t/biomarker-evaluation-c-statistic-auc-and-alternatives/6956/41 "2026-03-26T12:17:50Z")

</div>

> [@Uriah](#):
>
> Honestly the only performance metrics I care about from the “Discrimination” category are PPV and Lift - but they also hardly related to patient-specific decision making and more about organization-resource-constraint type of decision making.

Uriah these are generalized by distributions of predicted risk, back-to-back histograms, and related scatter plots as exemplified [here](https://fharrell.com/post/addvalue).

---

<div class="post-metadata">

**Author:** ![Uriah](https://discourse.datamethods.org/user_avatar/discourse.datamethods.org/uriah/32/1280_2.png) [@Uriah](https://discourse.datamethods.org/u/Uriah)\
**Post date:** [March 26, 2026, 12:55pm UTC](https://discourse.datamethods.org/t/biomarker-evaluation-c-statistic-auc-and-alternatives/6956/42 "2026-03-26T12:55:59Z")

</div>

I feel you 🙂

I think it’s part of our duty as “methodology folks” to fight the weird tendency to overcomplicate simple solutions. As @f2harrell suggests, I bet most physicians can see the point of back-to-back histograms of pre-post test distributions.

My wife just gave birth to my first child and a huge challenge was to fight the urge to do billions of unnecessary tests that contribute zero information and huge amount of anxiety.

It’s fairly tempting for clinicians to scare you with false-positive, the price is paid mainly by the patients. This discussion is impossible to make without understanding information gained by diagnosis and/or probability threshold as an exchange rate for decision making.

---

<div class="post-metadata">

**Author:** ![JiaqiLi](https://discourse.datamethods.org/letter_avatar_proxy/v4/letter/j/74df32/32.png) [@JiaqiLi](https://discourse.datamethods.org/u/JiaqiLi)\
**Post date:** [March 27, 2026, 5:41am UTC](https://discourse.datamethods.org/t/biomarker-evaluation-c-statistic-auc-and-alternatives/6956/43 "2026-03-27T05:41:25Z")

</div>

Completely agree.

And congratulations on the birth of your child, it is truly one of the most beautiful moments in life.🎉

---

<div class="post-metadata">

**Author:** ![Raj](https://discourse.datamethods.org/user_avatar/discourse.datamethods.org/raj/32/169_2.png) [@Raj](https://discourse.datamethods.org/u/Raj)\
**Post date:** [March 30, 2026, 2:08pm UTC](https://discourse.datamethods.org/t/biomarker-evaluation-c-statistic-auc-and-alternatives/6956/44 "2026-03-30T14:08:07Z")

</div>

thank you for this post. I have been reading about NRI, and I have found it very confusing. I dont really understand why AHA is justifying the use of NRI, which critically informs several of their recommendations in the new AHA lipid guidelines. I was unsure if it was perhaps some new or novel accepted statisticall approach to categorical or event rate NRI, but it seems a poor choice compared to alternatives.

---

<div class="post-metadata">

**Author:** ![f2harrell](https://discourse.datamethods.org/user_avatar/discourse.datamethods.org/f2harrell/32/165_2.png) [@f2harrell](https://discourse.datamethods.org/u/f2harrell)\
**Post date:** [March 30, 2026, 7:52pm UTC](https://discourse.datamethods.org/t/biomarker-evaluation-c-statistic-auc-and-alternatives/6956/45 "2026-03-30T19:52:10Z")

</div>

See Margaret Pepe’s and others’ severe criticisms of NRI.

[Previous page](https://discourse.datamethods.org/t/biomarker-evaluation-c-statistic-auc-and-alternatives/6956.md?page=2)