Anxi-Insom-Net: A Novel Omnipresent Machine Learning-Based Fortress Service Model to Predict Insomnia and Anxiety Disorder and Their Co-Occurrence
Year: 2024
Type: article
Study design: Observational Study
Abstract: The purpose of this research was to develop a service model for detecting anxiety and insomnia within a secure cloud environment and to determine whether there is a potential link between these conditions, such that an individual with anxiety may also have insomnia or vice versa. In this research, we aim to address the problem of detecting Insomnia and Anxiety disorder by constructing a new hybrid model. A survey was conducted in Bangladesh using both online and offline methods. A total of 3013 responses were gathered from different areas of the country using well-established medical questionnaires - HAM-A (Hamilton Anxiety Rating Scale) consisting of 14 questions to assess anxiety levels and ISI-7 (Insomnia Severity Index) consisting of 7 questions to evaluate insomnia. To predict mental illness in the country, five renowned machine learning algorithms were used: K-nearest neighbours, support vector machine (SVM), random forest (RF), Naive Bayes and linear discriminate analysis. Before and after applying the algorithms, statistical analysis and pre-processing of the dataset were carried out. The study also calculated the statistical correlation to determine the likelihood of someone experiencing insomnia if they had an anxiety disorder and vice versa. The SVM algorithm is the most accurate in detecting anxiety, with a 93.5% success rate and also in detecting insomnia, with a 94.36% success rate. This paper proposes a new hybrid model that can detect anxiety disorders with an accuracy of 98.67% and insomnia with an accuracy of 95.36%. Furthermore, this paper discovers that a person with an anxiety disorder has a 90.52% chance of also having insomnia, while a person with insomnia has a 93% chance of having an anxiety disorder. This study discovered that SVM offers higher accuracy in handling these types of datasets. Furthermore, our proposed model offers even greater accuracy than SVM. Additionally, individuals who experience anxiety are at a high risk of developing insomnia and vice versa.
Language: English
Subfield: Health Information Management
Field: Health Professions
Domain: Health Sciences
Open Access status: hybrid
OpenAlex ID: W4408481822