Receiver Operating Characteristic (ROC) Curve Analysis for Optimal Cut-off in Disease Identification
Вставка
- Опубліковано 7 сер 2024
- This video presents how to perform ROC analysis in determining the optimal cut-off for disease identification. It starts with the manual calculations of identifying the measures of accuracy (sensitivity, specificity, etc) for each cut-off, sketching the ROC curve, and eventually using Youden's index J for the optimal cut-off. The use of R was demonstrated on how to generate the same results, and for Area Under the Curve (AUC) calculation.
Data may be downloaded at drive.google.com/open?id=1rh5....
Video Chapters:
0:00 Introduction
2:59 Manual calculation and concepts using example 1 (miRNA.RData)
25:20 Using R (miRNA.RData)
31:59 Example 2 (elastase.RData)
34:50 Reminders in using ROC Curve
This is the best explanation and I really appreciate this. It is just fantastic !!!!!!
One of the best demonstration of cut-off values on UA-cam. 👍👍👍
Very well explained. Step-by-step and easy to understand. Thank you
This is the best explanation of Co and ROC I came across
Very good basic math and statistics foundation combined with good examples
Many thanks
Thank you so much. It's amazing. You saved my life..
Totally outstanding.....I can't explain...how much it helps me...thanks a lot.
You’re very welcome. I am glad to be of help :)
Thanks so much for the very clear explanation!
Great explanation. 👍
Thank you so much for this video. You help me a lot in writing my proposal. It is going to help me a lot in my project. May God bless you abundantly. Gracias
Superb and ultimate explanation. Thankyou
thank you so much!! You explained this so clearly, understandably and thoroughly! thanks a lot
Outstanding
Excellent
You are a gem
Anyway it’s excellent to learn and understand underpinning knowledge in ROC and cut off determinations
suberb explanation. thanks
Thank you very much for this great explanation!
I wondered if there is a possibility to extract the information from the plot (roc-curve) into a table. So that i can see the sens and spec for all possibile cut-offs. I coudn't find anything about this in the r package handbook. Do you have an idea, how i could transform this into a table?
Are you a Filipino sir? Got curious with the accent hehe. This helped me a lot. Thank you!
yes po ☺️
Thanks for the video, nice explanation. Could you please explain why the values of J in ID's 19 and 20 are neglected? They have the highest J indices...
I'm sorry I meant the J indices at 17:39 min. ID 19 has J = 0.798 and ID 20 has J = 0.909. Still, ID 12 with J = 0.616 is chosen as the cut off.
@@NishantSewgobind Oh that one. In ID 19, Sens = 0.0909 and Spec = 0.889, so J index is 0.0901 + 0.889 - 1 = -0.02 (not 0.798), while for ID 20, Sens = 0.0909 + Spec = 1, so J index is 0.0909 (not 0.909) . Im sorry about that but i think there was a typo error on those two in the powerpoint i used. THANKS A LOT for raising that concern. I didn't see that. Nonetheless, the Max J index is still the one at ID 12, which J index is 0.616.
@@xanmos aha! Thanks for the explanation, really appreciated 😄👍
Hello! This topic is quite far away from my specification so maybe I have stupid question. It’s about ECLIA test for Covid and the results are in COI, where COI1 is reactive and positive for antibodies. I just dont understand if the higher number tells me something about quantity of antibodies or no. Basically I’m just curious what that number tells. Brother was tested for Covid antibodies and his number is COI=11. Thanks for your answer.
I am not a medical laboratory scientist to answer ur question perfectly but as far as my readings are concerned (as to what i understood), yes the cutoff index (COI) indicates the amount of antibodies (specifically nucleocapsid) produced while patients have COVID-19. The nucleocapsid is at its peak on the first 12hrs of infection so maybe ur brother is at its early infection state when ELICA test was performed thats why it’s high to as much as 11.
you may refer to these articles:
mdpi-res.com/d_attachment/diagnostics/diagnostics-11-01808/article_deploy/diagnostics-11-01808.pdf
www.rrh.org/documents/COVID-19-Ab-Fact-Sheet.pdf
Thank u.
Can i share that SPSS can now calculate younden’s index or J stat
Used SPSS V29
thank u. I just got spss trial version and yes, there’s youden index already 🥳
could you demonstrate how to add two graph curve?
It can be done using the pROC package in R or RStudio (not using optimalCutoff package). Thank you for watching and appreciating my work. :)
Good day. I would like to ask for your help. If I want to calculate the sensitivity and specifity for a different cut-off point, how do you run it in R using the same package? Thank you for your help in advance
Hi, you have example in Python? Thanks!
Sorry i do not use Python.
@@xanmos Hi, have you done simultaneously optimizing cut-points for more than 1 variable or joint dichotomization?
thank you so much , I was wondering if you have any idea how find these by Graphpad prism
Medcalc does this technique, but i think Graphpad does not.
@@xanmos thank you so much for the reply, could you please share a link to your video if you made just the section when you did it on the R program.
@@linanaji6184 in 25:22 of this video, I started demonstrating in R.
@@xanmos I cannot download the OptimalCutpoints pcakage
@@linanaji6184 once R is installed, just enter
install.packages("OptimalCutpoints")
followed by
library(OptimalCutpoints)