An Intelligent Psychiatric Recommendation System for Detecting Mental Disorders

EasyChair Preprint no. 5341, version 1

Abstract

Nowadays, despite all the remarkable developments in mental healthcare, there are many uncertainties in the diagnosis process. Even psychiatric interviews with detailed, well-timed, and good follow-up systems may not be sufficient for differential diagnosis. At the same time, the workload of specialists increases in this diagnostic effort and it becomes very difficult to receive medical services and manage the treatment process due to the insufficient number of specialists. These problems, increase the need for auxiliary systems that will help experts in the diagnosis, save labor, and time. For this reason, we proposed a new intelligent psychiatric recommendation system with the Comprehensive Psychiatric Differential Diagnosis Test(CPDDT), which we created to screen-differentiate psychiatric diagnoses. To guide the expert in the system, in addition to the axis one and axis two diagnosis groups that refer to clinical disorders and personality disorders in DSM-4, it was aimed to measure the areas that affect the course of the illness and the treatment plan of the specialist, such as functionality, memory or suicidal thoughts. Thus, CPDDT was created, which could detect 48 different diagnostic groups in 319 questions. The test then was applied online to 676 users via a web system produced by DNB Analytics and psychiatrists evaluated the results in the clinic. Afterward, the test results were then evaluated by the Evolutionary Simulation Annealing LASSO Logistic Regression. As a result of this algorithm, after determining the importance of each question in the scale, the questions with low impact were eliminated and the test was reduced to 147 questions with .93 accuracy. In addition, the algorithm also found the probability of each patient being sick. In summary, the new machine-learning-based CPDDT was finalized with the number of 147 questions and the algorithm was presented as a suggestion system to the diagnostic process of experts.

Keyphrases: Differential Diagnosis, feature selection, machine learning, Mental Health Disorders, Psychiatric Diagnosis, Recommendation Systems

@Booklet{EasyChair:5341,
  author = {Esma Nur Ucar and Sedat Irgil and Salih Tutun and Nilay Aras and Ilker Yesilkaya},
  title = {An Intelligent Psychiatric Recommendation System for Detecting Mental Disorders},
  howpublished = {EasyChair Preprint no. 5341},

  year = {EasyChair, 2021}}