Week 2: Modifying Gieia’s PHQ-9 scoring to validate Gieia’s accurate assessment of teens’ mental wellness

Apr 09, 2021

This week I worked on modifying Gieia’s PHQ-9 scoring based on the contrast that I previously noted between the manually-computed scores and the Gieia-produced scores. I began by comparing the manual PHQ-9 scoring guide to Gieia’s algorithm. However, I found that the logic of the scoring guide and Gieia’s algorithm were identical. As a result, I decided to explore a different potential root of this issue.

I recalled that in the PHQ-9’s scoring guide, each selected answer is associated with a specific numerical value. Thus I decided to check whether the numerical value that each answer within Gieia was associated with matched the PHQ-9’s scoring guide. To check this, I modified the code such that every time I selected an answer for each question, I could see what numerical value this answer was associated with and how this answer had altered the cumulative score compiled for Gieia’s PHQ-9 thus far. Based on this, I learned that each answer in Gieia’s PHQ-9 was associated with a numerical value that was one greater than it should have been (as stipulated by the PHQ-9 scoring guide). So I altered the sector of the code assigning the numerical values associated with each answer choice accordingly.

To ensure that Gieia’s PHQ-9 scoring was now accurately representing the scoring guide, I tested a few of the manual PHQ-9 datasets that I had previously created within Gieia’s PHQ-9. However, I found that though some of the Gieia-produced scores and the manually-produced scores matched, some did not.

As a result, I consulted the modulations that I had previously made to the code such that every time I selected an answer for each question, I saw what numerical value this answer was associated with. These modulations also allowed me to view how this selected answer had altered the cumulative score compiled for Gieia’s PHQ-9 thus far. Based on this, I learned that Gieia was not adding the numerical value associated with the answer selected for the PHQ-9’s last question to the cumulative score, thus skewing the Gieia-produced PHQ-9 score and diagnosis. So I adjusted the code to include this final numerical value in the calculated cumulative score. 

After again testing a few of the manual PHQ-9 datasets within Gieia’s PHQ-9 to ensure that Gieia’s scoring was now accurately representing the scoring guide, I was able to confirm that Gieia is able to accurately assess teens’ mental wellness. 

2 Replies to “Week 2: Modifying Gieia’s PHQ-9 scoring to validate Gieia’s accurate assessment of teens’ mental wellness”

  1. Alex G. says:

    It seems your project is moving in a right way! One question I have is how would you ensure the validity of the scoring guide of Gieia would accurately illustrate the client’s mental status?

    1. Kamie A. says:

      Hi Alex! The scoring guides that I’m using within Gieia have been validated either in published research or by the medical organizations that developed the corresponding screening tools so they are able to accurately depict teens’ mental state.

Leave a Reply