Background of Machine/Deep Learning Approaches on Mental Health: History
View Latest Version
Please note this is an old version of this entry, which may differ significantly from the current revision.
Contributor:
Ngumimi Karen Iyortsuun
,
Soo-Hyung Kim
,
Min Jhon
,
Hyung-Jeong Yang
,
Sudarshan Pant

Mental health can be seen as a person’s emotional, psychological, and social well-being. It can be harmed by various mental health conditions, which negatively influence a person’s intellectual capacity, emotions, and social relationships. Machine learning (ML) is a subfield of artificial intelligence (AI) that deals with three problems: classification, regression, and clustering. It utilizes data and algorithms to mimic how people learn while progressively improving accuracy in various tasks.

  • machine learning
  • deep learning
  • mental health conditions

1. Introduction

Mental health can be seen as a person’s emotional, psychological, and social well-being. It can be harmed by various mental health conditions, which negatively influence a person’s intellectual capacity, emotions, and social relationships. To combat these disorders, appropriate and timely assessment is essential to identify (diagnose) one from the other. The screening of mental health conditions is performed using self-report questionnaires designed to detect certain sensations or attitudes toward social interactions [1].
Machine learning (ML) is a subfield of artificial intelligence (AI) that deals with three problems: classification, regression, and clustering. It utilizes data and algorithms to mimic how people learn while progressively improving accuracy in various tasks [2]. ML has been applied to multiple areas of psychological treatments and offers excellent potential for predicting and treating mental health conditions and analogous health outcomes. Typically, these algorithms require significant data to learn patterns and perform classification tasks. One of the most widely applied ML approaches in the prediction of mental illnesses is supervised learning.
Supervised learning is the process of learning a mapping of a collection of input variables and an output variable and applying this mapping to predict the outcomes of unseen data [3]. Support vector machine (SVM) is a good example of supervised learning that deals with classification and regression problems. This method works based on the concept of margin calculation by finding the optimal decision line or boundary called the hyperplane to separate n-dimensional space into different classes. This involves placing new data points into the correct categories in the future. Some advantages of SVM include its ability to handle both semi-structured and structured data. Additionally, because it adopts generalization, there is a lower probability of overfitting. However, SVM also has some disadvantages. With large datasets, there is an increase in training time. Therefore, its performance begins to dwindle. Additionally, SVM does not work well on a noisy dataset. Decision trees are also supervised learning methods for classification and regression problems. A tree can be seen as a piecewise constant approximation. It creates models that predict the value of target variables by learning simple decision rules inferred from data features. Logistic regression predicts the output of a categorical dependent variable; therefore, its outcome can either be Yes or No, 0 or 1, etc. Finally, naïve Bayes uses the Bayes theorem of probability for classification. It assumes that a particular feature is unrelated to other features in a given dataset.
Another widely used ML approach is ensemble learning. It involves training several individual learners to solve a problem. This method creates multiple learners and combines them to form a single model, and each learner works as an individual traditional ML method. Ensemble learning comprises three classes, bagging, boosting, and stacking. Bagging creates multiple datasets through random sampling, builds multiple learners in parallel, and combines all the learners using an average or majority vote strategy. Boosting creates multiple datasets through random sampling with replacement overweighted data and builds learners sequentially. These learners are then combined using a weighted averaging strategy. Stacking, on the other hand, either begins with bagging or boosting, and the outputs of the learners serve as inputs to another traditional ML algorithm (meta-model). The meta-model then acts as an aggregate by combining the outputs to provide results. Random Forest (RF) and extreme gradient boosting (XGBoost) are some of the most widely used ensemble learning methods. Random forest uses the bagging method to create decision trees with subsets of data, and each decision tree’s output is combined to make a final decision tree. XGBoost, on the other hand, is a scalable distributed gradient-boosting method of the decision tree.
Transfer learning is another ML method that researchers in this area are exploring. In simple terms, it is the transfer of knowledge from a related task that has already been learned to improve learning in a new task [4]. Although these algorithms open up new avenues for psychological research [5], their widespread use raises ethical and legal concerns about data anonymization.
ML is divided into various subfields, one of which is deep learning (DL) [6]. DL is a branch of ML that can take unstructured data such as text and images in its raw form and automatically finds the set of characteristics that differentiate distinct categories of data. Hence, one does not need to identify features as the architecture learns these features and increments on its own; therefore, it requires the utilization of a more extensive amount of data. Recently, there has been much interest in developing DL for mental illness diagnosis.
Since the introduction of AI into the medical sector, numerous studies and publications have been conducted on the use of ML and DL to intensify the examination of different medical problems. The application of AI in the medical sector has also extended to mental health condition diagnosis due to its great importance [7]. A number of advancements have been made in the application of ML for the diagnosis of mental health conditions. Integration with electronic health records (EHRs) is one such advancement. It is a growing trend in analyzing data from EHRs to assist with diagnosing mental health conditions. ML algorithms are also trained to analyze data retrieved from wearable devices such as smartwatches and fitness trackers. This approach has the potential to provide continuous monitoring of mental health status and enable the early detection of potential issues. Additionally, ML is applied in predictive modeling. Here, ML algorithms can identify individuals at risk of developing mental health conditions. This can allow for early intervention and prevent more severe mental health issues. Finally, ML is applied in the development of automated screening tools to help identify individuals at risk for certain mental health conditions.
ML is used to identify mental health conditions by analyzing patterns in data indicative of certain conditions. These data can be generated and collected from various sources, such as patient records, brain imaging scans, or even social media posts. For this purpose, different algorithms are used, including supervised learning algorithms, which are trained on labeled data, and unsupervised learning algorithms, which can identify patterns in data without the need for explicit labels. Once a model has been trained on the collected dataset, it can then be used to predict the likelihood that an individual has a particular mental health condition based on their data. ML researchers perform this prediction by applying the learned patterns to new data and using the models’ output to make a diagnosis.
Research in this area has been carried out using various ML techniques, and recently has been noted to extend to DL. In [8], Shamshirband et al. examined the use of convolutional neural networks (CNN), deep belief networks (DBN), auto-encoders (AE), and recurrent neural networks (RNN) in healthcare systems. They addressed several concerns and challenges with DL models in healthcare, as well as significant insights into the accuracy and applicability of DL models. In another review [9], the authors focused on previous studies on ML to predict general mental health problems and proposed possible future avenues for investigation.
Librenza-Garcia et al. [10] reviewed past studies on diagnosing bipolar disorder patients through ML techniques. He et al. [11] surveyed automatic depression estimation (ADE) methods relating to deep neural networks (DNN) and presented architectures based on audio-visual cues. Finally, in a review of PTSD, Ramos-Lima [12] reviewed the use of ML techniques in assessing subjects with PTSD and acute stress disorder (ASD).

2. Background

According to the World Health Organization (WHO), in 2019, anxiety and depression were the most common mental health conditions among the estimated 970 million people worldwide living with mental health problems. However, this number rose remarkably due to the onset of the COVID-19 pandemic in 2020. With this pandemic grew the importance of gaining access to medical care and treating mental illnesses. Although these options exist, most people do not gain access to them, and many of them face discrimination, stigma, and violation of their human rights [13].
Diagnosing mental health issues involves a thorough psychiatric interview, usually covering the suspected symptoms, psychiatric history, and physical examinations. Psychological tests and assessment tools are also helpful when identifying psychiatric symptoms [14].
Schizophrenia is a severe mental illness that affects a person’s ability to interpret reality, thus causing an abnormal interpretation of reality. A report by the World Health Organization stated that schizophrenia affects at least 1 in 300 people worldwide. Additionally, it increases the likeliness of death of patients by about two to three times due to their proneness to cardiovascular, metabolic, and infectious diseases [15]. It may result in delusions, hallucinations, disorganized speech, disorganized behavior, and negative symptoms. This may ultimately lead to social and occupational dysfunction [16].
Depression (major depressive disorder) is one of the widespread mental illness commonly screened through the Patient Health Questionnaire (PHQ) [17]. It is usually identified through symptoms of deep sadness and loss of interest in activities, leading to a decrease in a person’s ability to function correctly. Shorey et al. [18], in their study, stated that 34% of adolescents between the ages of 10 and 19 are at risk of clinical depression, exceeding the estimates of individuals aged between 18 and 25. Their study also showed that the Middle East, Africa, and Asia have the highest prevalence of elevated depressive symptoms; however, female adolescents reportedly have a higher prevalence of elevated depressive symptoms than male adolescents. Depression, if not properly attended to, may result in suicidal ideations and suicide [19].
Anxiety brings about the feeling of worry or fear that can be mild or severe. Anxiety on its own is a symptom of several other conditions, such as social anxiety disorder (social phobia), panic disorder, and phobias. Although everyone feels some anxiety at some point, it becomes a problem to be taken into serious consideration when they find it hard to control these feelings when they constantly affect their daily lives. Some general anxiety symptoms include dizziness or heart palpitations, trouble sleeping, a lack of concentration, restlessness, and worry. It is estimated that about 264 million people suffer from anxiety disorder, and a study conducted in 2020 showed that 62% of respondents to a survey reported some degree of anxiety, and a higher percentage of those affected by this disorder are women [20].
Another form of mental disorder is called bipolar disorder, formally known as “manic-depressive illness” or “manic-depression,” which causes an unusual change in mood, a reduction in energy, and lower activity levels and concentration levels. The stated mood changes range from periods of extreme highs (manic episodes) to lows (depressive episodes), as well as less severe episodes (hypomanic episodes). Studies show that about 46 million people worldwide present with bipolar disorder at different levels. People with this disorder may also be at risk of suicide, with about 60% showing signs of substance misuse [21]. It is usually diagnosed at different points in a person’s life based on symptoms, life history, experiences, and on rare occasions, family history.
Flashbacks, nightmares, and severe anxiety characterize PTSD, as well as constant uncontrollable thoughts triggered by terrifying events that a person either experienced or witnessed. To properly diagnose PTSD, medical personnel perform physical examinations on the suspected patient to check for medical issues that may have caused the prevailing symptoms. They conduct a psychological evaluation to discuss the events that might have triggered the appearance of the symptoms and use the criteria in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) to diagnose the illness efficiently [13]. Like most mental illnesses, PTSD is also not curable, but can be managed with proper treatment (mostly psychotherapy), which can help an affected person gain control over their life. With a lifetime prevalence of 8% in adolescents between the ages of 13 and 18, statistics also show that about 3.5% of U.S. adults report cases of PTSD yearly [22].
Anorexia nervosa is a life-threatening eating disorder with no fully recognized etiology that affects people of all ages, regardless of gender. Statistics show that about 9% of the population worldwide suffer from eating disorders and about 26% of those affected are at risk of suicide attempts [23]. Over the years, anorexia has become the most dangerous mental health condition among young women and girls in well-developed societies [24]. The affected people place extreme importance on controlling their body weight and shape using severe methods. They are usually never satisfied with their body weight, no matter how much weight they lose. They may drastically control their calorie intake by throwing up after eating or misusing laxatives, diuretics, or enemas. They may also exercise excessively to lose weight [25].
Finally, when considering neurodevelopmental disorders, two major conditions stand out: autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD). ASD impacts development and involves a broad range of conditions characterized by challenges with social skills, restricted and repetitive behaviors, and deficits in speech and nonverbal communication. ASD and ADHD are both neurodevelopmental disorders that tend to have neurological effects on the functioning of the nervous system, with similar symptoms. Often, one disorder is confused with the other. Children with ADHD usually have a lot of trouble focusing. They tend to be overly active and act without control over their impulses. ADHD is a mental illness that runs in families and is hard to be cured, although it can be managed if diagnosed earlier in the child’s life. Several steps are involved in diagnosing ADHD. The suspected patient must have shown about six or more inattentiveness, hyperactivity, and impulsiveness symptoms. The family history of the suspected patient is also taken into consideration. ADHD is treated with a combination of medication and behavior therapy.

This entry is adapted from the peer-reviewed paper 10.3390/healthcare11030285

References

  1. Hamilton, M. Development of a Rating Scale for Primary Depressive Illness. Br. J. Soc. Clin. Psychol. 1967, 6, 278–296.
  2. Mitchell, T.M. Machine Learning; McGraw-Hill: New York, NY, USA, 1997; Volume 1.
  3. Cunningham, P.; Cord, M.; Delany, S. Supervised Learning. In Machine Learning Techniques for Multimedia; Springer: Berlin/Heidelberg, Germany, 2008; pp. 21–49.
  4. Torrey, L.; Shavlik, J. Transfer Learning. In Handbook of Research on Machine Learning Applications and Trends: Algorithms, Methods, and Techniques; IGI Global: Hershey, PA, USA, 2010; pp. 242–264.
  5. Wongkoblap, A.; Vadillo, M.A.; Curcin, V. Researching Mental Health Disorders in the Era of Social Media: Systematic Review. J. Med. Internet Res. 2017, 19, e228.
  6. LeCun, Y.; Bengio, Y.; Hinton, G. Deep learning. Nature 2015, 521, 436–444.
  7. Graham, S.; Depp, C.; Lee, E.E.; Nebeker, C.; Tu, X.; Kim, H.-C.; Jeste, D.V. Artificial Intelligence for Mental Health and Mental Illnesses: An Overview. Curr. Psychiatry Rep. 2019, 21, 116.
  8. Shamshirband, S.; Fathi, M.; Dehzangi, A.; Chronopoulos, A.T.; Alinejad-Rokny, H. A review on deep learning approaches in healthcare systems: Taxonomies, challenges, and open issues. J. Biomed. Inform. 2020, 113, 103627.
  9. Chung, J.; Teo, J. Mental Health Prediction Using Machine Learning: Taxonomy, Applications, and Challenges. Appl. Comput. Intell. Soft Comput. 2022, 2022, 9970363.
  10. Librenza-Garcia, D.; Kotzian, B.J.; Yang, J.; Mwangi, B.; Cao, B.; Lima, L.N.P.; Bermudez, M.B.; Boeira, M.V.; Kapczinski, F.; Passos, I.C. The impact of machine learning techniques in the study of bipolar disorder: A systematic review. Neurosci. Biobehav. Rev. 2017, 80, 538–554.
  11. He, L.; Niu, M.; Tiwari, P.; Marttinen, P.; Su, R.; Jiang, J.; Guo, C.; Wang, H.; Ding, S.; Wang, Z.; et al. Deep learning for depression recognition with audiovisual cues: A review. Inf. Fusion 2021, 80, 56–86.
  12. Ramos-Lima, L.F.; Waikamp, V.; Antonelli-Salgado, T.; Passos, I.C.; Freitas, L.H.M. The use of machine learning techniques in trauma-related disorders: A systematic review. J. Psychiatr. Res. 2020, 121, 159–172.
  13. WHO. Mental Disorders. 2022. Available online: https://www.who.int/news-room/fact-sheets/detail/mental-disorders (accessed on 18 August 2022).
  14. Jencks, S.F. Recognition of mental distress and diagnosis of mental disorder in primary care. JAMA 1985, 253, 1903–1907.
  15. Schizophrenia. Available online: https://www.who.int/news-room/fact-sheets/detail/schizophrenia (accessed on 27 December 2022).
  16. Patel, K.R.; Cherian, J.; Gohil, K.; Atkinson, D. Schizophrenia: Overview and treatment options. Peer Rev. J. Formul. Manag. 2014, 39, 638–645.
  17. Costantini, L.; Pasquarella, C.; Odone, A.; Colucci, M.E.; Costanza, A.; Serafini, G.; Aguglia, A.; Murri, M.B.; Brakoulias, V.; Amore, M.; et al. Screening for depression in primary care with Patient Health Questionnaire-9 (PHQ-9): A systematic review. J. Affect. Disord. 2020, 279, 473–483.
  18. Shorey, S.; Ng, E.D.; Wong, C.H.J. Global prevalence of depression and elevated depressive symptoms among adolescents: A systematic review and meta-analysis. Br. J. Clin. Psychol. 2021, 61, 287–305.
  19. Harmer, B.; Lee, S.; Duong, T.V.H.; Saadabadi, A. Suicidal Ideation. In StatPearls; StatPearls Publishing: Treasure Island, FL, USA, 2022.
  20. SingleCare, T. Anxiety Statistics 2022. 2022. Available online: https://www.singlecare.com/blog/news/anxiety-statistics/ (accessed on 27 December 2022).
  21. SingleCare, T. Bipolar Disorder Statistics 2022. 2022. Available online: https://www.singlecare.com/blog/news/bipolar-disorder-statistics/ (accessed on 27 December 2022).
  22. Taylor-Desir, M. What Is Posttraumatic Stress Disorder (PTSD)? 2022. Available online: https://www.psychiatry.org/patients-families/ptsd/what-is-ptsd (accessed on 27 December 2022).
  23. Anad. Eating Disorder Statistics. 2021. Available online: https://anad.org/eating-disorders-statistics/ (accessed on 27 December 2022).
  24. Spinczyk, D.; Bas, M.; Dzieciątko, M.; Maćkowski, M.; Rojewska, K.; Maćkowska, S. Computer-aided therapeutic diagnosis for anorexia. Biomed. Eng. Online 2020, 19, 53.
  25. Clinic, M. Anorexia Nervosa. 1998–2022. Available online: https://www.mayoclinic.org/diseases-conditions/anorexia-nervosa/symptoms-causes/syc-20353591 (accessed on 17 August 2022).
More
© Text is available under the terms and conditions of the Creative Commons Attribution (CC BY) license; additional terms may apply. By using this site, you agree to the Terms and Conditions and Privacy Policy.
This entry is offline, you can click here to edit this entry!
Video Production Service
Feedback