TY  - JOUR
PY  - 2024//
TI  - Prediction of suicidal ideation with associated risk factors among university students in the southern part of Bangladesh: machine learning approach
JO  - Journal of affective disorders
A1  - Sara, Sabiha Shirin
A1  - Rahman, Md Asikur
A1  - Rahman, Riaz
A1  - Talukder, Ashis
SP  - ePub
EP  - ePub
VL  - ePub
IS  - ePub
N2  - BACKGROUND: The prevalence of suicidal ideation has become an urgent issue, particularly among adolescents. The primary objective of this research is to determine the prevalence of suicidal ideation among students in the southern region of Bangladesh and to predict this phenomenon using machine learning (ML) models. <br><br>METHODS: The data collection process involved using a simple random sampling technique to gather information from university students located in the southern region of Bangladesh during the period spreading from April 2022 to June 2022. Upon accounting for missing values and non-response rates, the ultimate sample size was determined to be 584, with 51.5 % of participants identifying as male and 48.5 % female. <br><br>RESULTS: A significant proportion of students, precisely 19.9 %, reported experiencing suicidal ideation. Most participants were female (77 %) and unmarried (78 %). Within the machine learning (ML) framework, KNN exhibited the highest accuracy score of 91.45 %. In addition, the Random Forest (RF), and Categorical Boosting (CatBoost) algorithms exhibited comparable levels of accuracy, achieving scores of 90.60 and 90.59 respectively. LIMITATIONS: Using a cross-sectional design in research limits the ability to establish causal relationships. <br><br>CONCLUSION: Mental health practitioners can employ the KNN model alongside patients' medical histories to detect those who may be at a higher risk of attempting suicide. This approach enables healthcare professionals to take appropriate measures, such as counselling, encouraging regular sleep patterns, and addressing depression and anxiety, to prevent suicide attempts.<p />  <p>Language: en</p>
LA  - en
SN  - 0165-0327
UR  - http://dx.doi.org/10.1016/j.jad.2024.01.092
ID  - ref1
ER  -