Loading...

Performance Evaluation of Machine Learning and Deep Learning Algorithms in Psychiatric Disorders Classification Using the RDoC Clinical Dataset

Ehyaei, Zeinab | 2024

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 57942 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Manzuri Shalmani, Mohammad Taghi
  7. Abstract:
  8. Today, psychiatric disorders have turned into a widespread epidemic worldwide. The World Health Organization estimates that one in every four individuals experiences a mental disorder at least once during their lifetime. Mental health problems not only directly affect the health of the affected individuals but also impose significant social and economic costs on the health system and society in terms of diagnosing and treating these disorders. Early, accurate diagnosis and effective treatment of mental health-related disorders can alleviate the suffering of individuals grappling with these illnesses. Machine learning methods are employed as a desirable solution for identifying mental health issues. This research focuses on examining a novel dataset from a machine learning perspective, designed and collected by a team of professors at Tehran University of Medical Sciences. This dataset is structured based on the Research Domain Criteria (RDoC) and is being used for the first time. In this study, the performance of traditional machine learning methods and a category of deep learning methods adapted for the classification of tabular data is evaluated. Additionally, the impact of utilizing deep generative networks to generate additional artificial datasets on the learning power of the model in the RDoC system is investigated. The results of this research indicate that compared to the Diagnostic and Statistical Manual of Psychiatric, the use of RDoC features increases the accuracy of classifying psychiatric patients. XGBoost, CatBoost, and AutoInt algorithms have shown better performance in classifying this dataset. Furthermore, the results indicate that using deep generative models for generating more artificial datasets leads to an increase in the model’s AUC-ROC in the RDoC system
  9. Keywords:
  10. Machine Learning ; Deep Learning ; Mental Disorders Diagnostic and Statistical Manual ; Research Domain Criteria ; Psychiatric Disorders Classification

 Digital Object List

 Bookmark

No TOC