Name: Design of a predictive model of people´s mental health state based on music consumption patterns
Uploaded: 2025-01-26
Duration: 3 min 28 s
Description: Design of a predictive model of people´s mental health state based on music consumption patterns.

Design of a predictive model of people´s mental health state based on music consumption patterns. 

Scene 1

Design of a predictive model of people´s mental health state based on music consumption patterns

Carlos Moreno (UCV), Lindsay Dietz(Shion Consulting), Bianca Vilchez (Taxes DR) and Julian Arias(UPM).

"AI & Its Applications : Transforming Industries and Society"

A1 & Its Application
Ttmstormlna InaustrlSB arid SOCIetv
UAE

Biography:-
-Electronics Engineering (USB,VZLA)
-Specialist on Data Network (UCV,VZLA)
-M.Sc Data Networks (UCV)
-PhD. Telecommunications (Paris VI / TelecomSudParis, France)
-PostDoc on Data Engineering (UPM, Madrid)
-Master on Big Data & BI (NextEd., Madrid)
-MBA (IESA, VZLA)
-Full Professor (Titular, UCV). 

Scene 2

Biography:-
-Electronics Engineering (USB,VZLA)
-Specialist on Data Network (UCV,VZLA)
-M.Sc Data Networks (UCV)
-PhD. Telecommunications (Paris VI / TelecomSudParis, France)
-PostDoc on Data Engineering (UPM, Madrid)
-Master on Big Data & BI (NextEd., Madrid)
-MBA (IESA, VZLA)
-Full Professor (Titular, UCV)

•••,fau.tnor q...umsay .tnoå a.'01dx"d

Abstract:-. "AI & Its Applications : Transforming Industries and Society". 

Scene 3

This research project attempts to design a predictive model to analyze the mental health status reported by individuals and based on their music consumption patterns. In this work, two predictive machine learning submodels were designed based on the Random Forest algorithm, taking as input variables the attributes of the two datasets (control and disorder) obtained from public repositories that contained information of tweets from a large group of anonymous users, such as song, author, listening time, mental health status and platform.  4 variations of  Random Forest training were used for classifier A (binary output) or for classifier B (multiclass). The hyperparameters were number of trees and depth. The results show that it may be possible to predict the mental health states reported by the group of users.

State of the Art:-. "AI & Its Applications : Transforming Industries and Society". 

Scene 4

(Lacárcel,  2003)  and (García & Fernández-Company, 2023). Terapeutical characteristics of music.
 Data Science and	Big Data in	sociology and psychology (Becerra, 2018)

Music and Artificial Intelligence (Huang & Huang, 2020).
 (Li & Li, 2022) Extraction of variables to  categorize genres and musical styles.
 (Zheng & Lam, 2022) Effect of music on students.
 Studies are not sufficient on detecting disorder from music consumption, by using Big Data and AI.

Hypothesis:  It is possible to detect the presence of mental health disorder and relationship with music consumption patterns

Description:-. "AI & Its Applications : Transforming Industries and Society". 

Scene 5

*ilFERP
Explore Your Research Journey...

One-hot-encoding (disorder,src, Genres,part_of_day)

- Accuracy - Standard Deviation - Error - ROC

- Accuracy Standard Deviation -    Error -    ROC -   Confusion  Matrix

- System with A and B      - Confusion Matrix      - Sensitivity Recall

A1. RF with Cross-Validation A2. RF with Stratified K-Fold A3. RF with  Balanced Training A4. RF with  SMOTE B1. RF with Cross-Validation B2. RF with Stratified K-Fold B3. RF with  Balanced Training B4. RF with  SMOTE

Conclusion:. "AI & Its Applications : Transforming Industries and Society". 

Scene 6

-  Preprocessing and transformation of the data involved significant computational complexity The conformation of the final dataset for this research process took the greatest amount of time and effort The best A and B models were designed based on the cross-validation technique with SMOTE in the sampling using Random Forest with a moderate number of trees and a predetermined tree depth The general model requires significant improvements in order to become a more robust and accurate predictive model for Anxiety and Depression. It is possible to detect the presence of mental health disorder and relationship with music consumption patterns.

Appendix:. "AI & Its Applications : Transforming Industries and Society". 

Scene 7

•••fatunor pnasaH .tnoå a.'01d.r"t1

asu
969
0/09
0/08
0/0LL
0/09
0/08L
ssvO IPOH
uo!ssaxdaa
0/001
0/09'
0/009
O/oz
0/09
0/089
0/08'
0/091
0/08
0/06
O/oc
Yoc
Åp!xuv
loxW0D
u oæssaxdaa
asu

General Model  Recall (Sensitivity):
 Control             53% Depression      63% Anxiety             74% Other                82% PTSD                 77%

Best A Model: A4 with 100 trees default depth. Accuracy 64% AUC 0.72 Best B Model: B4 with 40 trees default depth. Accuracy 33% AUC 0.56

Thanking you…!!. Thanks for your assistance…!! carlosmanuel.moreno.ucv@gmail.com + 58-414-9191033 Chief of Unit of Networks and Measurements School of Electrical Engineering Central University of Venezuela. 

Scene 8

Thanks for your assistance…!!
 carlosmanuel.moreno.ucv@gmail.com
 + 58-414-9191033  Chief of Unit of Networks and Measurements School of Electrical Engineering Central University of Venezuela

Design of a predictive model of people´s mental health state based on music consumption patterns

Scene 1 (0s)

Scene 2 (16s)

Scene 3 (36s)

Scene 4 (1m 10s)

Scene 5 (1m 37s)

Scene 6 (2m 20s)

Scene 7 (2m 46s)

Scene 8 (3m 13s)