Machine Learning Approaches for Clinical Psychology and Psychiatry

Suicide attempts and non-suicidal self-injury in Chinese adolescents: Predictive models using a neural network model

INTRODUCTION

Suicide attempts (SA) are a significant contributor to suicide deaths, and non-suicidal self-injury (NSSI) can increase the risk of SA. Many adolescents experience both NSSI and SA, which are affected by various factors. This study aimed to identify the risk factors and essential warning signs of SA, establish a predictive model for SA using multiple dimensions and large samples, and provide a multidimensional perspective for clinical diagnosis and intervention.

METHODS

A total of 9140 participants aged 12-18 years participated in an online survey; 6959 participants were included in the statistical analysis. A multilayer perceptron algorithm was used to establish a prediction model for adolescent SA (with or without); adolescents with NSSI behavior were extracted as a subgroup to establish a prediction model.

RESULTS

Both the prediction model performance of the SA group and the NSSI-SA subgroup were strong, with high accuracy, and AUC values of 0.93 and 0.88, indicating good discrimination. Decision curve analysis (DCA) demonstrated that the clinical intervention value of the prediction results was high and that the clinical intervention benefits of the NSSI-SA subgroup were greater than those of the SA group.

CONCLUSIONS

Our study demonstrated that the predictive model has a high degree of accuracy and discrimination, thereby identifying significant factors associated with adolescent SA. As long as adolescents exhibit NSSI behavior, relative suicide interventions should be implemented to prevent future hazards. This study can provide guidance and more nuanced insights for clinical diagnosis as well as a foundation for clinical treatment.

Spatiotemporal models of dengue epidemiology in the Philippines: Integrating remote sensing and interpretable machine learning

Previous dengue epidemiological analyses have been limited in spatiotemporal extent or covariate dimensions, the latter neglecting the multifactorial nature of dengue. These constraints, caused by rigid and traditional statistical tools which collapse amidst 'Big Data', prompt interpretable machine-learning (iML) approaches. Predicting dengue incidence and mortality in the Philippines, a data-limited yet high-burden country, the mlr3 universe of R packages was used to build and optimize ML models based on remotely sensed provincial and dekadal 3 NDVI and 9 rainfall features from 2016 to 2020. Between two tasks, models differ across four random forest-based learners and two clustering strategies. Among 16 candidates, rfsrc-year-case and ranger-year-death significantly perform best for predicting dengue incidence and mortality, respectively. Therefore, temporal clustering yields the best models, reflective of dengue seasonality. The two best models were subjected to tripartite global exploratory model analyses, which encompass model-agnostic post-hoc methods such as Permutation Feature Importance (PFI) and Accumulated Local Effects (ALE). PFI reveals that the models differ in their important explanatory aspect, rainfall for rfsrc-year-case and NDVI for ranger-year-death, among which long-term average (lta) features are most relevant. Trend-wise, ALE reveals that average incidence predictions are positively associated with 'Rain.lta', reflective of dengue cases peaking during the wet season. In contrast, those for mortality are negatively associated with 'NDVI.lta', reflective of urban spaces driving dengue-related deaths. By technologically addressing the challenges of the human-animal-ecosystem interface, this study adheres to the One Digital Health paradigm operationalized under Sustainable Development Goals (SDGs). Leveraging data digitization and predictive modeling for epidemiological research paves SDG 3, which prioritizes holistic health and well-being.

Navigating merits and limits on the current perspectives and ethical challenges in the utilization of artificial intelligence in psychiatry - An exploratory mixed methods study

BACKGROUND

The integration of Artificial Intelligence (AI) in psychiatry presents opportunities for enhancing patient care but raises significant ethical concerns and challenges in clinical application. Addressing these challenges necessitates an informed and ethically aware psychiatric workforce capable of integrating AI into practice responsibly.

METHODS

A mixed-methods study was conducted to assess the outcomes of the "CONNECT with AI" - (Collaborative Opportunity to Navigate and Negotiate Ethical Challenges and Trials with Artificial Intelligence) workshop, aimed at exploring AI's ethical implications and applications in psychiatry. This workshop featured presentations, discussions, and scenario analyses focusing on AI's role in mental health care. Pre- and post-workshop questionnaires and focus group discussions evaluated participants' perspectives, and ethical understanding regarding AI in psychiatry.

RESULTS

Participants exhibited a cautious optimism towards AI, recognizing its potential to augment mental health care while expressing concerns over ethical usage, patient-doctor relationships, and AI's practical application in patient care. The workshop significantly improved participants' ethical understanding, highlighting a substantial knowledge gap and the need for further education in AI among psychiatrists.

CONCLUSION

The study underscores the necessity of continuous education and ethical guideline development for psychiatrists in the era of AI, emphasizing collaborative efforts in AI system design to ensure they meet clinical needs ethically and effectively. Future initiatives should aim to broaden psychiatrists' exposure to AI, fostering a deeper understanding and integration of AI technologies in psychiatric practice.

Diagnostic value of regional homogeneity and fractional amplitude of low-frequency fluctuations in the classification of schizophrenia and bipolar disorders

Schizophrenia (SCZ) and bipolar disorders (BD) show significant neurobiological and clinical overlap. In this study, we wanted to identify indexes of intrinsic brain activity that could differentiate these disorders. We compared the diagnostic value of the fractional amplitude of low-frequency fluctuations (fALFF) and regional homogeneity (ReHo) estimated from resting-state functional magnetic resonance imaging in a support vector machine classification of 59 healthy controls (HC), 40 individuals with SCZ, and 43 individuals with BD type I. The best performance, measured by balanced accuracy (BAC) for binary classification relative to HC was achieved by a stacking model (87.4% and 90.6% for SCZ and BD, respectively), with ReHo performing better than fALFF, both in SCZ (86.2% vs. 79.4%) and BD (89.9% vs. 76.9%). BD were better differentiated from HC by fronto-temporal ReHo and striato-temporo-thalamic fALFF. SCZ were better classified from HC using fronto-temporal-cerebellar ReHo and insulo-tempo-parietal-cerebellar fALFF. In conclusion, we provided evidence of widespread aberrancies of spontaneous activity and local connectivity in SCZ and BD, demonstrating that ReHo features exhibited superior discriminatory power compared to fALFF and achieved higher classification accuracies. Our results support the complementarity of these measures in the classification of SCZ and BD and suggest the potential for multivariate integration to improve diagnostic precision.

Use of Machine-Learning Algorithms Based on Text, Audio and Video Data in the Prediction of Anxiety and Post-Traumatic Stress in General and Clinical Populations: A Systematic Review

Research in machine-learning (ML) algorithms using natural behavior (i.e., text, audio, and video data) suggests that these techniques could contribute to personalization in psychology and psychiatry. However, a systematic review of the current state-of-the-art is missing. Moreover, individual studies often target ML experts, and may overlook potential clinical implications of their findings. In a narrative accessible to mental health professionals, we present a systematic review, conducted in 5 psychology and 2 computer-science databases. We included 128 studies assessing the predictive power of ML algorithms using text, audio, and/or video data in the prediction of anxiety and post-traumatic stress (PTSD). Most studies (n = 87) aimed at predicting anxiety, the remainder (n = 41) focused on PTSD. They were mostly published since 2019, in computer science journals, and tested algorithms using text (n = 72), as opposed to audio or video. They focused mainly on general populations (n = 92), less on laboratory experiments (n = 23) or clinical populations (n = 13). Methodological quality varied, as did reported metrics of the predictive power, hampering comparison across studies. Two thirds of studies, focusing on both disorders, reported acceptable to very good predictive power (including high-quality studies only). Results of 33 studies were uninterpretable, mainly due to missing information. Research into ML algorithms using natural behavior is in its infancy, but shows potential to contribute to diagnostics of mental disorders, such as anxiety and PTSD, in the future, if standardization of methods, reporting of results, and research in clinical populations are improved.

Machine learning identifies different related factors associated with depression and suicidal ideation in Chinese children and adolescents

BACKGROUND

Depression and suicidal ideation often co-occur in children and adolescents, yet they possess distinct characteristics. This study sought to identify the different related factors associated with depression and suicidal ideation.

METHODS

A nationwide cross-sectional survey collected data from Chinese children and adolescents aged 8 to 18 (N = 160,962; 48.91 % girls). The survey included inquiries about demographics, depression, suicidal ideation, anxiety, perceived stress, academic burnout, internet addiction, non-suicidal self-injury, bullying, and being bullied. Fifteen machine learning algorithms were conducted to identify the different related factors associated with depression and suicidal ideation. Additionally, we conducted external validation on an independent sample of 1,812,889 children and adolescents.

RESULTS

Our findings revealed seven related factors linked to depression and six associated with suicidal ideation, with average accuracy rates of 86.86 % and 85.82 %, respectively. For depression, the most influential factors were anxiety, perceived stress, academic burnout, internet addiction, non-suicidal self-injury, experience of bullying, and age. Similarly, anxiety, non-suicidal self-injury, perceived stress, internet addiction, academic burnout, and age emerged as paramount factors for suicidal ideation. Moreover, these related factors showed notable variations in their predictive capacities for depression and suicidal ideation across different subgroups.

CONCLUSION

Anxiety emerged as the predominant shared factor for both depression and suicidal ideation, whereas the other related factors displayed distinct predictive patterns for each condition. These findings highlight the critical need for tailored strategies from public mental health service providers and policymakers to address the pressing concerns of depression and suicidal ideation.

Machine learning based on functional and structural connectivity in mild cognitive impairment

OBJECTIVE

Alzheimer's disease (AD) is a chronic, degenerative neurological disorder characterized by progressive cognitive decline and mental behavioral abnormalities. Mild cognitive impairment (MCI) is regarded as a transitional stage in the progression from normal elderly individuals to patients with AD. While studies have identified abnormalities in brain connectivity in patients with MCI, including functional and structural connectivity, accurately identifying patients with MCI in clinical screening remains challenging. We hypothesized that utilizing machine learning (ML) based on both functional and structural connectivity could yield meaningful results in distinguishing between patients with MCI and normal elderly individuals, so as to provide valuable information for early diagnosis and precise evaluation of patients with MCI.

METHODS

Following clinical criteria, we recruited 32 patients with MCI for the patient group, and 32 normal elderly individuals for the control group. All subjects underwent examinations for resting-state functional magnetic resonance imaging (rs-fMRI) and diffusion tensor imaging (DTI). Subsequently, significant functional and structural connectivity features were selected and combined with a support vector machine for classification of the patient and control groups.

RESULTS

We observed significantly different functional connectivity in the frontal lobe and putamen between the MCI group and normal controls. The results based on functional connectivity features demonstrated a classification accuracy of 71.88% and an area under the curve (AUC) value of 0.78. In terms of structural connectivity, we found that decreased fractional anisotropy in patients with MCI was significantly associated with Montreal Cognitive Assessment scores, specifically in regions such as the precuneus and cingulate gyrus. The classification results using the structural connectivity feature yielded an accuracy of 92.19% and an AUC value of 0.99. Lastly, combining functional and structural connectivity features resulted in a classification accuracy and AUC value of 93.75% and 0.99, respectively.

CONCLUSIONS

In this study, we demonstrated a high classification performance, underscoring the potential of both brain functional and structural connectivity in distinguishing patients with MCI from normal elderly individuals. Furthermore, the integration of functional connectivity and structural connectivity features indicated that utilizing rs-fMRI and DTI could enhance the accuracy and specificity of identifying patients with MCI compared with relying on a single neuroimaging technique.

Precision psychiatry: predicting predictability

Precision psychiatry is an emerging field that aims to provide individualized approaches to mental health care. An important strategy to achieve this precision is to reduce uncertainty about prognosis and treatment response. Multivariate analysis and machine learning are used to create outcome prediction models based on clinical data such as demographics, symptom assessments, genetic information, and brain imaging. While much emphasis has been placed on technical innovation, the complex and varied nature of mental health presents significant challenges to the successful implementation of these models. From this perspective, I review ten challenges in the field of precision psychiatry, including the need for studies on real-world populations and realistic clinical outcome definitions, and consideration of treatment-related factors such as placebo effects and non-adherence to prescriptions. Fairness, prospective validation in comparison to current practice and implementation studies of prediction models are other key issues that are currently understudied. A shift is proposed from retrospective studies based on linear and static concepts of disease towards prospective research that considers the importance of contextual factors and the dynamic and complex nature of mental health.

Enhancing psychiatric rehabilitation outcomes through a multimodal multitask learning model based on BERT and TabNet: An approach for personalized treatment and improved decision-making

Evaluating the rehabilitation status of individuals with serious mental illnesses (SMI) necessitates a comprehensive analysis of multimodal data, including unstructured text records and structured diagnostic data. However, progress in the effective assessment of rehabilitation status remains limited. Our study develops a deep learning model integrating Bidirectional Encoder Representations from Transformers (BERT) and TabNet through a late fusion strategy to enhance rehabilitation prediction, including referral risk, dangerous behaviors, self-awareness, and medication adherence, in patients with SMI. BERT processes unstructured textual data, such as doctor's notes, whereas TabNet manages structured diagnostic information. The model's interpretability function serves to assist healthcare professionals in understanding the model's predictive decisions, improving patient care. Our model exhibited excellent predictive performance for all four tasks, with an accuracy exceeding 0.78 and an area under the curve of 0.70. In addition, a series of tests proved the model's robustness, fairness, and interpretability. This study combines multimodal and multitask learning strategies into a model and applies it to rehabilitation assessment tasks, offering a promising new tool that can be seamlessly integrated with the clinical workflow to support the provision of optimized patient care.

Predicting post-treatment symptom severity for adults receiving psychological therapy in routine care for generalised anxiety disorder: a machine learning approach

Approximately half of generalised anxiety disorder (GAD) patients do not recover from first-line treatments, and no validated prediction models exist to inform individuals or clinicians of potential treatment benefits. This study aimed to develop and validate an accurate and explainable prediction model of post-treatment GAD symptom severity. Data from adults receiving treatment for GAD in eight Improving Access to Psychological Therapies (IAPT) services (n=15,859) were separated into training, validation and holdout datasets. Thirteen machine learning algorithms were compared using 10-fold cross-validation, against two simple clinically relevant comparison models. The best-performing model was tested on the holdout dataset and model-specific explainability measures identified the most important predictors. A Bayesian Additive Regression Trees model out-performed all comparison models (MSE=16.54 [95 % CI=15.58; 17.51]; MAE=3.19; R²=0.33, including a single predictor linear regression model: MSE=20.70 [95 % CI=19.58; 21.82]; MAE=3.94; R²=0.14). The five most important predictors were: PHQ-9 anhedonia, GAD-7 annoyance/irritability, restlessness and fear items, then the referral-assessment waiting time. The best-performing model accurately predicted post-treatment GAD symptom severity using only pre-treatment data, outperforming comparison models that approximated clinical judgement and remaining within the GAD-7 error of measurement and minimal clinically important differences. This model could inform treatment decision-making and provide desired information to clinicians and patients receiving treatment for GAD.

Predicting life satisfaction using machine learning and explainable AI

Life satisfaction is a crucial facet of human well-being. Hence, research on life satisfaction is incumbent for understanding how individuals experience their lives and influencing interventions targeted at enhancing mental health and well-being. Life satisfaction has traditionally been measured using analog, complicated, and frequently error-prone methods. These methods raise questions concerning validation and propagation. However, this study demonstrates the potential for machine learning algorithms to predict life satisfaction with a high accuracy of 93.80% and a 73.00% macro F1-score. The dataset comes from a government survey of 19000 people aged 16-64 years in Denmark. Using feature learning techniques, 27 significant questions for assessing contentment were extracted, making the study highly reproducible, simple, and easily interpretable. Furthermore, clinical and biomedical large language models (LLMs) were explored for predicting life satisfaction by converting tabular data into natural language sentences through mapping and adding meaningful counterparts, achieving an accuracy of 93.74% and macro F1-score of 73.21%. It was found that life satisfaction prediction is more closely related to the biomedical domain than the clinical domain. Ablation studies were also conducted to understand the impact of data resampling and feature selection techniques on model performance. Moreover, the correlation between primary determinants with different age brackets was analyzed, and it was found that health condition is the most important determinant across all ages. The best performing Machine Learning model trained in this study is deployed on a public server, ensuring unrestricted usage of the model. We highlight the advantages of machine learning methods for predicting life satisfaction and the significance of XAI for interpreting and validating these predictions. This study demonstrates how machine learning, large language models and XAI can jointly contribute to building trust and understanding in using AI to investigate human behavior, with significant ramifications for academics and professionals working to quantify and comprehend subjective well-being.

A machine learning model to predict the risk of perinatal depression: Psychosocial and sleep-related factors in the Life-ON study cohort

Perinatal depression (PND) is a common complication of pregnancy associated with serious health consequences for both mothers and their babies. Identifying risk factors for PND is key to early detect women at increased risk of developing this condition. We applied a machine learning (ML) approach to data from a multicenter cohort study on sleep and mood changes during the perinatal period ("Life-ON") to derive models for PND risk prediction in a cross-validation setting. A wide range of sociodemographic variables, blood-based biomarkers, sleep, medical, and psychological data collected from 439 pregnant women, as well as polysomnographic parameters recorded from 353 women, were considered for model building. These covariates were correlated with the risk of future depression, as assessed by regularly administering the Edinburgh Postnatal Depression Scale across the perinatal period. The ML model indicated the mood status of pregnant women in the first trimester, previous depressive episodes and marital status, as the most important predictors of PND. Sleep quality, insomnia symptoms, age, previous miscarriages, and stressful life events also added to the model performance. Besides other predictors, sleep changes during early pregnancy should therefore assessed to identify women at higher risk of PND and support them with appropriate therapeutic strategies.

Optimizing Outcomes in Psychotherapy for Anxiety Disorders Using Smartphone-Based and Passive Sensing Features: Protocol for a Randomized Controlled Trial

BACKGROUND

Psychotherapies, such as cognitive behavioral therapy (CBT), currently have the strongest evidence of durable symptom changes for most psychological disorders, such as anxiety disorders. Nevertheless, only about half of individuals treated with CBT benefit from it. Predictive algorithms, including digital assessments and passive sensing features, could better identify patients who would benefit from CBT, and thus, improve treatment choices.

OBJECTIVE

This study aims to establish predictive features that forecast responses to transdiagnostic CBT in anxiety disorders and to investigate key mechanisms underlying treatment responses.

METHODS

This study is a 2-armed randomized controlled clinical trial. We include patients with anxiety disorders who are randomized to either a transdiagnostic CBT group or a waitlist (referred to as WAIT). We index key features to predict responses prior to starting treatment using subjective self-report questionnaires, experimental tasks, biological samples, ecological momentary assessments, activity tracking, and smartphone-based passive sensing to derive a multimodal feature set for predictive modeling. Additional assessments take place weekly at mid- and posttreatment and at 6- and 12-month follow-ups to index anxiety and depression symptom severity. We aim to include 150 patients, randomized to CBT versus WAIT at a 3:1 ratio. The data set will be subject to full feature and important features selected by minimal redundancy and maximal relevance feature selection and then fed into machine leaning models, including eXtreme gradient boosting, pattern recognition network, and k-nearest neighbors to forecast treatment response. The performance of the developed models will be evaluated. In addition to predictive modeling, we will test specific mechanistic hypotheses (eg, association between self-efficacy, daily symptoms obtained using ecological momentary assessments, and treatment response) to elucidate mechanisms underlying treatment response.

RESULTS

The trial is now completed. It was approved by the Cantonal Ethics Committee, Zurich. The results will be disseminated through publications in scientific peer-reviewed journals and conference presentations.

CONCLUSIONS

The aim of this trial is to improve current CBT treatment by precise forecasting of treatment response and by understanding and potentially augmenting underpinning mechanisms and personalizing treatment.

TRIAL REGISTRATION

ClinicalTrials.gov NCT03945617; https://clinicaltrials.gov/ct2/show/results/NCT03945617.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

DERR1-10.2196/42547.

Machine learning approach for the development of a crucial tool in suicide prevention: The Suicide Crisis Inventory-2 (SCI-2) Short Form

The Suicide Crisis Syndrome (SCS) describes a suicidal mental state marked by entrapment, affective disturbance, loss of cognitive control, hyperarousal, and social withdrawal that has predictive capacity for near-term suicidal behavior. The Suicide Crisis Inventory-2 (SCI-2), a reliable clinical tool that assesses SCS, lacks a short form for use in clinical settings which we sought to address with statistical analysis. To address this need, a community sample of 10,357 participants responded to an anonymous survey after which predictive performance for suicidal ideation (SI) and SI with preparatory behavior (SI-P) was measured using logistic regression, random forest, and gradient boosting algorithms. Four-fold cross-validation was used to split the dataset in 1,000 iterations. We compared rankings to the SCI-Short Form to inform the short form of the SCI-2. Logistic regression performed best in every analysis. The SI results were used to build the SCI-2-Short Form (SCI-2-SF) utilizing the two top ranking items from each SCS criterion. SHAP analysis of the SCI-2 resulted in meaningful rankings of its items. The SCI-2-SF, derived from these rankings, will be tested for predictive validity and utility in future studies.

Delusions and Delinquencies: A Comparison of Violent and Non-Violent Offenders With Schizophrenia Spectrum Disorders

The relationship between schizophrenia spectrum disorders (SSD) and violent offending has long been the subject of research. The present study attempts to identify the content of delusions, an understudied factor in this regard, that differentiates between violent and non-violent offenses. Limitations, clinical relevance, and future directions are discussed. Employing a retrospective study design, machine learning algorithms and a comprehensive set of variables were applied to a sample of 366 offenders with a schizophrenia spectrum disorder in a Swiss forensic psychiatry department. Taking into account the different contents and affects associated with delusions, eight variables were identified as having an impact on discriminating between violent and non-violent offenses with an AUC of 0.68, a sensitivity of 30.8%, and a specificity of 91.9%, suggesting that the variables found are useful for discriminating between violent and non-violent offenses. Delusions of grandiosity, delusional police and/or army pursuit, delusional perceived physical and/or mental injury, and delusions of control or passivity were more predictive of non-violent offenses, while delusions with aggressive content or delusions associated with the emotions of anger, distress, or agitation were more frequently associated with violent offenses. Our findings extend and confirm current research on the content of delusions in patients with SSD. In particular, we found that the symptoms of threat/control override (TCO) do not directly lead to violent behavior but are mediated by other variables such as anger. Notably, delusions traditionally seen as symptoms of TCO, appear to have a protective value against violent behavior. These findings will hopefully help to reduce the stigma commonly and erroneously associated with mental illness, while supporting the development of effective therapeutic approaches.

Characterizing cognitive subtypes in schizophrenia using cortical curvature

Cognitive deficits are a core symptom of schizophrenia, but research on their neural underpinnings has been challenged by the heterogeneity in deficits' severity among patients. Here, we address this issue by combining logistic regression and random forest to classify two neuropsychological profiles of patients with high (HighCog) and low (LowCog) cognitive performance in two independent samples. We based our analysis on the cortical features grey matter volume (VOL), cortical thickness (CT), and mean curvature (MC) of N = 57 patients (discovery sample) and validated the classification in an independent sample (N = 52). We investigated which cortical feature would yield the best classification results and expected that the 10 most important features would include frontal and temporal brain regions. The model based on MC had the best performance with area under the curve (AUC) values of 76% and 73%, and identified fronto-temporal and occipital brain regions as the most important features for the classification. Moreover, subsequent comparison analyses could reveal significant differences in MC of single brain regions between the two cognitive profiles. The present study suggests MC as a promising neuroanatomical parameter for characterizing schizophrenia cognitive subtypes.

Leveraging Novel Technologies and Artificial Intelligence to Advance Practice-Oriented Research

Mental health services are experiencing notable transformations as innovative technologies and artificial intelligence (AI) are increasingly utilized in a growing number of studies and services.These cutting-edge technologies carry the promise of substantial improvements in the field of mental health. Nevertheless, questions emerge about the alignment of novel technologies and AI systems with human needs, especially in the context of vulnerable populations receiving mental healthcare. The practice-oriented research (POR) model is pivotal in seamlessly integrating these emerging technologies into clinical research and practice. It underscores the importance of tight collaboration between clinicians and researchers, all driven by the central goal of ensuring and elevating client well-being. This paper focuses on how novel technologies can enhance the POR model and highlights its pivotal role in integrating these technologies into clinical research and practice. We discuss two key phases: pre-treatment, and during treatment. For each phase, we describe the challenges, present the major technological innovations, describe recent studies exemplifying technology use, and suggest future directions. Ethical concerns and the importance of aligning humans and technology are also considered, in addition to implications for practice and training.

On the Selection of Item Scores or Composite Scores for Clinical Prediction

Recent shifts to prioritize prediction, rather than explanation, in psychological science have increased applications of predictive modeling methods. However, composite predictors, such as sum scores, are still commonly used in practice. The motivations behind composite test scores are largely intertwined with reducing the influence of measurement error in answering explanatory questions. But this may be detrimental for predictive aims. The present paper examines the impact of utilizing composite or item-level predictors in linear regression. A mathematical examination of the bias-variance decomposition of prediction error in the presence of measurement error is provided. It is shown that prediction bias, which may be exacerbated by composite scoring, drives prediction error for linear regression. This may be particularly salient when composite scores are comprised of heterogeneous items such as in clinical scales where items correspond to symptoms. With sufficiently large training samples, the increased prediction variance associated with item scores becomes negligible even when composite scores are sufficient. Practical implications of predictor scoring are examined in an empirical example predicting suicidal ideation from various depression scales. Results show that item scores can markedly improve prediction particularly for symptom-based scales. Cross-validation methods can be used to empirically justify predictor scoring decisions.

An Interpretable Machine Learning Approach to Predict Sensory Processing Sensitivity Trait in Nursing Students

Sensory processing sensitivity (SPS) is a personality trait that makes certain individuals excessively sensitive to stimuli. People carrying this trait are defined as Highly Sensitive People (HSP). The SPS trait is notably prevalent among nursing students and nurse staff. Although there are HSP diagnostic tools, there is little information about early detection. Therefore, the aim of this work was to develop a prediction model to identify HSP and provide an individualized nursing assessment. A total of 672 nursing students completed all the evaluations. In addition to the HSP diagnosis, emotional intelligence, communication skills, and conflict styles were evaluated. An interpretable machine learning model was trained to predict the SPS trait. We observed a 33% prevalence of HSP, which was higher in women and people with previous health training. HSP were characterized by greater emotional repair (p = 0.033), empathy (p = 0.030), respect (p = 0.038), and global communication skills (p = 0.036). Overall, sex and emotional intelligence dimensions are important to detect this trait, although personal characteristics should be considered. The present individualized prediction model could help to predict the presence of the SPS trait in nursing students, which may be useful in conducting intervention strategies to avoid the negative consequences and reinforce the positive ones of this trait.

Brief Report: Machine Learning for Estimating Prognosis of Children with Autism Receiving Early Behavioral Intervention-A Proof of Concept

Although early behavioral intervention is considered as empirically-supported for children with autism, estimating treatment prognosis is a challenge for practitioners. One potential solution is to use machine learning to guide the prediction of the response to intervention. Thus, our study compared five machine algorithms in estimating treatment prognosis on two outcomes (i.e., adaptive functioning and autistic symptoms) in children with autism receiving early behavioral intervention in a community setting. Each machine learning algorithm produced better predictions than random sampling on both outcomes. Those results indicate that machine learning is a promising approach to estimating prognosis in children with autism, but studies comparing these predictions with those produced by qualified practitioners remain necessary.

Cognitive and clinical predictors of a long-term course in obsessive compulsive disorder: A machine learning approach in a prospective cohort study

BACKGROUND

Obsessive compulsive disorder (OCD) is a disabling illness with a chronic course, yet data on long-term outcomes are scarce. This study aimed to examine the long-term course of OCD in patients treated with different approaches (drugs, psychotherapy, and psychosurgery) and to identify predictors of clinical outcome by machine learning.

METHOD

We included outpatients with OCD treated at our referral unit. Demographic and neuropsychological data were collected at baseline using standardized instruments. Clinical data were collected at baseline, 12 weeks after starting pharmacological treatment prescribed at study inclusion, and after follow-up.

RESULTS

Of the 60 outpatients included, with follow-up data available for 5-17 years (mean = 10.6 years), 40 (67.7 %) were considered non-responders to adequate treatment at the end of the study. The best machine learning model achieved a correlation of 0.63 for predicting the long-term Yale-Brown Obsessive Compulsive Scale (Y-BOCS) score by adding clinical response (to the first pharmacological treatment) to the baseline clinical and neuropsychological characteristics.

LIMITATIONS

Our main limitations were the sample size, modest in the context of traditional ML studies, and the sample composition, more representative of rather severe OCD cases than of patients from the general community.

CONCLUSIONS

Many patients with OCD showed persistent and disabling symptoms at the end of follow-up despite comprehensive treatment that could include medication, psychotherapy, and psychosurgery. Machine learning algorithms can predict the long-term course of OCD using clinical and cognitive information to optimize treatment options.

Clinical and neuroimaging predictors of benzodiazepine response in catatonia: A machine learning approach

Catatonia is a well characterized psychomotor syndrome combining motor, behavioural and neurovegetative signs. Benzodiazepines are the first-choice treatment, effective in 70 % of cases. Currently, the factors associated with benzodiazepine resistance remain unknown. We aimed to develop machine learning models using clinical and neuroimaging data to predict benzodiazepine response in catatonic patients. This study examined a cohort of catatonic patients who underwent standardized clinical evaluation, 3 T brain MRI, and benzodiazepine trial. Based on clinical response, patients were classified as benzodiazepine responders or non-responders. Cortical thickness and regional brain volumes were measured. Two machine learning models (linear model and gradient boosting tree model) were developed to identify predictors of treatment response using clinical, demographic, and neuroimaging data. The cohort included 65 catatonic patients, comprising 30 benzodiazepine responders and 35 non-responders. Using clinical data alone, the linear model achieved 63% precision, 51% recall, a specificity of 61%, and 58% AUC, while the gradient boosting tree (GBT) model attained 46% precision, 60% recall, a specificity of 62% and 64% AUC. Incorporating neuroimaging data improved model performance, with the linear model achieving 66% precision, 57% recall, a specificity of 67%, and 70% AUC, and the GBT model attaining 50% precision, 50% recall, a specificity of 62% and 70% AUC. The integration of imaging data with demographic and clinical information significantly enhanced the predictive performance of the models. The duration of the catatonic syndrome, along with the presence of mitgehen (passive obedience) and immobility/stupor, and the volume of the right medial orbito-frontal cortex emerged as important factors in predicting non-response to benzodiazepines.

Trajectories and predictors of response to social cognition training in people with schizophrenia: A proof-of-concept machine learning study

BACKGROUND

Social cognition training (SCT) can improve social cognition deficits in schizophrenia. However, little is known about patterns of response to SCT or individual characteristics that predict response.

METHODS

76 adults with schizophrenia randomized to receive 8-12 weeks of remotely-delivered SCT were included in this analysis. Social cognition was measured with a composite of six assessments. Latent class growth analyses identified trajectories of social cognitive response to SCT. Random forest and logistic regression models were trained to predict membership in the trajectory group that showed improvement from baseline measures including symptoms, functioning, motivation, and cognition.

RESULTS

Five trajectory groups were identified: Group 1 (29 %) began with slightly above average social cognition, and this ability significantly improved with SCT. Group 2 (9 %) had baseline social cognition approximately one standard deviation above the sample mean and did not improve with training. Groups 3 (18 %) and 4 (36 %) began with average to slightly below-average social cognition and showed non-significant trends toward improvement. Group 5 (8 %) began with social cognition approximately one standard deviation below the sample mean, and experienced significant deterioration in social cognition. The random forest model had the best performance, predicting Group 1 membership with an area under the curve of 0.73 (SD 0.24; 95 % CI [0.51-0.87]).

CONCLUSIONS

Findings suggest that there are distinct patterns of response to SCT in schizophrenia and that those with slightly above average social cognition at baseline may be most likely to experience gains. Results may inform future research seeking to individualize SCT treatment for schizophrenia.

Individualized prediction of anxiety and depressive symptoms using gray matter volume in a non-clinical population

Machine learning is an emerging tool in clinical psychology and neuroscience for the individualized prediction of psychiatric symptoms. However, its application in non-clinical populations is still in its infancy. Given the widespread morphological changes observed in psychiatric disorders, our study applies five supervised machine learning regression algorithms-ridge regression, support vector regression, partial least squares regression, least absolute shrinkage and selection operator regression, and Elastic-Net regression-to predict anxiety and depressive symptom scores. We base these predictions on the whole-brain gray matter volume in a large non-clinical sample (n = 425). Our results demonstrate that machine learning algorithms can effectively predict individual variability in anxiety and depressive symptoms, as measured by the Mood and Anxiety Symptoms Questionnaire. The most discriminative features contributing to the prediction models were primarily located in the prefrontal-parietal, temporal, visual, and sub-cortical regions (e.g. amygdala, hippocampus, and putamen). These regions showed distinct patterns for anxious arousal and high positive affect in three of the five models (partial least squares regression, support vector regression, and ridge regression). Importantly, these predictions were consistent across genders and robust to demographic variability (e.g. age, parental education, etc.). Our findings offer critical insights into the distinct brain morphological patterns underlying specific components of anxiety and depressive symptoms, supporting the existing tripartite theory from a neuroimaging perspective.

Predicting Lifetime Suicide Attempts in a Community Sample of Adolescents Using Machine Learning Algorithms

Suicide is a major global health concern and a prominent cause of death in adolescents. Previous research on suicide prediction has mainly focused on clinical or adult samples. To prevent suicides at an early stage, however, it is important to screen for risk factors in a community sample of adolescents. We compared the accuracy of logistic regressions, elastic net regressions, and gradient boosting machines in predicting suicide attempts by 17-year-olds in the Millennium Cohort Study (N = 7,347), combining a large set of self- and other-reported variables from different categories. Both machine learning algorithms outperformed logistic regressions and achieved similar balanced accuracies (.76 when using data 3 years before the self-reported lifetime suicide attempts and .85 when using data from the same measurement wave). We identified essential variables that should be considered when screening for suicidal behavior. Finally, we discuss the usefulness of complex machine learning models in suicide prediction.

Machine learning and the prediction of suicide in psychiatric populations: a systematic review

Machine learning (ML) has emerged as a promising tool to enhance suicidal prediction. However, as many large-sample studies mixed psychiatric and non-psychiatric populations, a formal psychiatric diagnosis emerged as a strong predictor of suicidal risk, overshadowing more subtle risk factors specific to distinct populations. To overcome this limitation, we conducted a systematic review of ML studies evaluating suicidal behaviors exclusively in psychiatric clinical populations. A systematic literature search was performed from inception through November 17, 2022 on PubMed, EMBASE, and Scopus following the PRISMA guidelines. Original research using ML techniques to assess the risk of suicide or predict suicide attempts in the psychiatric population were included. An assessment for bias risk was performed using the transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD) guidelines. About 1032 studies were retrieved, and 81 satisfied the inclusion criteria and were included for qualitative synthesis. Clinical and demographic features were the most frequently employed and random forest, support vector machine, and convolutional neural network performed better in terms of accuracy than other algorithms when directly compared. Despite heterogeneity in procedures, most studies reported an accuracy of 70% or greater based on features such as previous attempts, severity of the disorder, and pharmacological treatments. Although the evidence reported is promising, ML algorithms for suicidal prediction still present limitations, including the lack of neurobiological and imaging data and the lack of external validation samples. Overcoming these issues may lead to the development of models to adopt in clinical practice. Further research is warranted to boost a field that holds the potential to critically impact suicide mortality.

Disease Progression Patterns of Brain Morphology in Schizophrenia: More Progressed Stages in Treatment Resistance

BACKGROUND AND HYPOTHESIS

Given the heterogeneity and possible disease progression in schizophrenia, identifying the neurobiological subtypes and progression patterns in each patient may lead to novel biomarkers. Here, we adopted data-driven machine-learning techniques to identify the progression patterns of brain morphological changes in schizophrenia and investigate the association with treatment resistance.

STUDY DESIGN

In this cross-sectional multicenter study, we included 177 patients with schizophrenia, characterized by treatment response or resistance, with 3D T1-weighted magnetic resonance imaging. Cortical thickness and subcortical volumes calculated by FreeSurfer were converted into z scores using 73 healthy controls data. The Subtype and Stage Inference (SuStaIn) algorithm was used for unsupervised machine-learning analysis.

STUDY RESULTS

SuStaIn identified 3 different subtypes: (1) subcortical volume reduction (SC) type (73 patients), in which volume reduction of subcortical structures occurs first and moderate cortical thinning follows, (2) globus pallidus hypertrophy and cortical thinning (GP-CX) type (42 patients), in which globus pallidus hypertrophy initially occurs followed by progressive cortical thinning, and (3) cortical thinning (pure CX) type (39 patients), in which thinning of the insular and lateral temporal lobe cortices primarily happens. The remaining 23 patients were assigned to baseline stage of progression (no change). SuStaIn also found 84 stages of progression, and treatment-resistant schizophrenia showed significantly more progressed stages than treatment-responsive cases (P = .001). The GP-CX type presented earlier stages than the pure CX type (P = .009).

CONCLUSIONS

The brain morphological progressions in schizophrenia can be classified into 3 subtypes, and treatment resistance was associated with more progressed stages, which may suggest a novel biomarker.

Classifying autism in a clinical population based on motion synchrony: a proof-of-concept study using real-life diagnostic interviews

Predictive modeling strategies are increasingly studied as a means to overcome clinical bottlenecks in the diagnostic classification of autism spectrum disorder. However, while some findings are promising in the light of diagnostic marker research, many of these approaches lack the scalability for adequate and effective translation to everyday clinical practice. In this study, our aim was to explore the use of objective computer vision video analysis of real-world autism diagnostic interviews in a clinical sample of children and young individuals in the transition to adulthood to predict diagnosis. Specifically, we trained a support vector machine learning model on interpersonal synchrony data recorded in Autism Diagnostic Observation Schedule (ADOS-2) interviews of patient-clinician dyads. Our model was able to classify dyads involving an autistic patient (n = 56) with a balanced accuracy of 63.4% against dyads including a patient with other psychiatric diagnoses (n = 38). Further analyses revealed no significant associations between our classification metrics with clinical ratings. We argue that, given the above-chance performance of our classifier in a highly heterogeneous sample both in age and diagnosis, with few adjustments this highly scalable approach presents a viable route for future diagnostic marker research in autism.

Unveiling the potential of machine learning in schizophrenia diagnosis: A meta-analytic study of task-based neuroimaging data

The emergence of machine learning (ML) techniques has opened up new avenues for identifying biomarkers associated with schizophrenia (SCZ) using task-related fMRI (t-fMRI) designs. To evaluate the effectiveness of this approach, we conducted a comprehensive meta-analysis of 31 t-fMRI studies using a bivariate model. Our findings revealed a high overall sensitivity of 0.83 and specificity of 0.82 for t-fMRI studies. Notably, neuropsychological domains modulated the classification performance, with selective attention demonstrating a significantly higher specificity than working memory (β = 0.98, z = 2.11, P = 0.04). Studies involving older, chronic patients with SCZ reported higher sensitivity (P <0.015) and specificity (P <0.001) than those involving younger, first-episode patients or high-risk individuals for psychosis. Additionally, we found that the severity of negative symptoms was positively associated with the specificity of the classification model (β = 7.19, z = 2.20, P = 0.03). Taken together, these results support the potential of using task-based fMRI data in combination with machine learning techniques to identify biomarkers related to symptom outcomes in SCZ, providing a promising avenue for improving diagnostic accuracy and treatment efficacy. Future attempts to deploy ML classification should consider the factors of algorithm choice, data quality and quantity, as well as issues related to generalization.

Towards a taxonomy of mechanisms of change? Findings from an expert survey on the association between common factors and specific techniques in psychotherapy

OBJECTIVE

In the present study, we used structural equation modeling (SEM) to investigate the complex relationship between common factors, i.e., mechanisms of change, and specific factors, i.e., therapeutic techniques.

METHOD

N = 256 psychotherapy experts were asked to rate the appropriateness of 14 techniques commonly used in psychotherapy to facilitate five different common factors - resource activation, motivational clarification, self-management & emotion regulation, social competence, and therapeutic relationship. Using SEM, we defined techniques as indicators and common factors as latent variables. Data were split randomly into two subsets. Indicators were selected if three a priori defined criteria were met based on training data (n = 128). Subsequently, the goodness of model fit was assessed in the test data (n = 128).

RESULTS

The proposed model revealed adequate fit. All factor loadings were theoretically sound and significant in magnitude. Findings suggest that psychotherapy experts discriminate between common factors by their various associations with therapeutic techniques.

CONCLUSION

Suggestions are made, how therapeutic techniques are to be used to facilitate desirable change in the patient. Our model is a step towards a taxonomy of mechanisms of change that may help to improve research-informed decision-making.

[Diagnosis and admission center : Establishment and evaluation of an integrated translational infrastructure for clinical psychiatric research]

The routine in-depth characterization of patients with methods of clinical and scale-based examination, neuropsychology, based on biomaterials, and sensor-based information opens up transformative possibilities on the way to personalized diagnostics, treatment and prevention in psychiatry, psychotherapy, and psychosomatics. Effective integration of the additional temporal and logistical effort into everyday care as well as the acceptance by patients are critical to the success of such an approach but there is little evidence on this to date. We report here on the establishment of the Diagnosis and Admission Center (DAZ) at the Central Institute of Mental Health (ZI) in Mannheim. The DAZ is an outpatient unit upstream of other care structures for clinical and scientific phenotyping across diagnoses as a starting point for data-driven, individualized pathways to further treatment, diagnostics or research. We describe the functions, goals, and implementation of the newly created clinical scientific translational structure, provide an overview of the patient populations it has reached, and provide data on its acceptance. In this context, the close integration with downstream clinical processes enables a better coordinated and demand-oriented allocation. In addition, DAZ enables a faster start of disorder-specific diagnostics and treatment. Since its launch in April 2021 up to the end of 2022, 1021 patients underwent psychiatric evaluation at DAZ during a pilot phase. The patient sample corresponded to a representative sample from standard care and the newly established processes were regarded as helpful by patients. In summary, the DAZ uniquely combines the interests and needs of patient with the collection of scientifically relevant data.

Mobile-based ecological momentary assessment and intervention: bibliometric analysis

Objective

The aim of this article was to review existing documents in the field of mobile-based EMA and EMI, provide an overview of current hot topics, and predict future development trends.

Methods

We conducted a bibliometric study on mobile-based EMA and EMI publications that were collected from the Web of Science Core Collection database. Biblioshiny and CiteSpace were utilized to analyze scientific productions, leading sources, authors, affiliations, documents, research hot topics, keywords, and trend topics.

Results

A total of 2222 documents related to EMA and EMI published between 1992 and 2023 were retrieved. In recent years, scholarly publications have generally increased in mobile-based EMA and EMI research, particularly in the last decade. JMIR mHealth and uHealth (n=86), as well as JMIR (n=73), showed the highest number of publications. The United States (n=1038), Germany (n=218) and Netherlands (n=175) were leading countries. Regarding keyword co-occurrence and trend topics analysis, mental health, health behaviors, and feasibility were hot topics in mobile-based EMA and EMI research. Future research trends included using EMA for tailoring EMI, just-in-time adaptive interventions (JITAI), and digital phenotyping.

Conclusion

This bibliometric study on mobile-based EMA and EMI is a valuable resource for understanding the field's evolution and future trends. Our analysis indicates that EMA and EMI have great potential in health behaviors and mental health, but implementation should consider feasibility and reactivity issues carefully. Emerging trends include EMA-tailored EMI, JITAI, and digital phenotyping. In the future, strengthening multidisciplinary cooperation will be necessary to promote the continued development of the field.

Task-state skin potential abnormalities can distinguish major depressive disorder and bipolar depression from healthy controls

Early detection of bipolar depression (BPD) and major depressive disorder (MDD) has been challenging due to the lack of reliable and easily measurable biological markers. This study aimed to investigate the accuracy of discriminating patients with mood disorders from healthy controls based on task state skin potential characteristics and their correlation with individual indicators of oxidative stress. A total of 77 patients with BPD, 53 patients with MDD, and 79 healthy controls were recruited. A custom-made device, previously shown to be sufficiently accurate, was used to collect skin potential data during six emotion-inducing tasks involving video, pictorial, or textual stimuli. Blood indicators reflecting individual levels of oxidative stress were collected. A discriminant model based on the support vector machine (SVM) algorithm was constructed for discriminant analysis. MDD and BPD patients were found to have abnormal skin potential characteristics on most tasks. The accuracy of the SVM model built with SP features to discriminate MDD patients from healthy controls was 78% (sensitivity 78%, specificity 82%). The SVM model gave an accuracy of 59% (sensitivity 59%, specificity 79%) in classifying BPD patients, MDD patients, and healthy controls into three groups. Significant correlations were also found between oxidative stress indicators in the blood of patients and certain SP features. Patients with depression and bipolar depression have abnormalities in task-state skin potential that partially reflect the pathological mechanism of the illness, and the abnormalities are potential biological markers of affective disorders.

Machine learning classification of autism spectrum disorder based on reciprocity in naturalistic social interactions

Autism spectrum disorder is characterized by impaired social communication and interaction. As a neurodevelopmental disorder typically diagnosed during childhood, diagnosis in adulthood is preceded by a resource-heavy clinical assessment period. The ongoing developments in digital phenotyping give rise to novel opportunities within the screening and diagnostic process. Our aim was to quantify multiple non-verbal social interaction characteristics in autism and build diagnostic classification models independent of clinical ratings. We analyzed videos of naturalistic social interactions in a sample including 28 autistic and 60 non-autistic adults paired in dyads and engaging in two conversational tasks. We used existing open-source computer vision algorithms for objective annotation to extract information based on the synchrony of movement and facial expression. These were subsequently used as features in a support vector machine learning model to predict whether an individual was part of an autistic or non-autistic interaction dyad. The two prediction models based on reciprocal adaptation in facial movements, as well as individual amounts of head and body motion and facial expressiveness showed the highest precision (balanced accuracies: 79.5% and 68.8%, respectively), followed by models based on reciprocal coordination of head (balanced accuracy: 62.1%) and body (balanced accuracy: 56.7%) motion, as well as intrapersonal coordination processes (balanced accuracy: 44.2%). Combinations of these models did not increase overall predictive performance. Our work highlights the distinctive nature of non-verbal behavior in autism and its utility for digital phenotyping-based classification. Future research needs to both explore the performance of different prediction algorithms to reveal underlying mechanisms and interactions, as well as investigate the prospective generalizability and robustness of these algorithms in routine clinical care.

Functional magnetic resonance imaging in schizophrenia: current evidence, methodological advances, limitations and future directions

Functional neuroimaging emerged with great promise and has provided fundamental insights into the neurobiology of schizophrenia. However, it has faced challenges and criticisms, most notably a lack of clinical translation. This paper provides a comprehensive review and critical summary of the literature on functional neuroimaging, in particular functional magnetic resonance imaging (fMRI), in schizophrenia. We begin by reviewing research on fMRI biomarkers in schizophrenia and the clinical high risk phase through a historical lens, moving from case-control regional brain activation to global connectivity and advanced analytical approaches, and more recent machine learning algorithms to identify predictive neuroimaging features. Findings from fMRI studies of negative symptoms as well as of neurocognitive and social cognitive deficits are then reviewed. Functional neural markers of these symptoms and deficits may represent promising treatment targets in schizophrenia. Next, we summarize fMRI research related to antipsychotic medication, psychotherapy and psychosocial interventions, and neurostimulation, including treatment response and resistance, therapeutic mechanisms, and treatment targeting. We also review the utility of fMRI and data-driven approaches to dissect the heterogeneity of schizophrenia, moving beyond case-control comparisons, as well as methodological considerations and advances, including consortia and precision fMRI. Lastly, limitations and future directions of research in the field are discussed. Our comprehensive review suggests that, in order for fMRI to be clinically useful in the care of patients with schizophrenia, research should address potentially actionable clinical decisions that are routine in schizophrenia treatment, such as which antipsychotic should be prescribed or whether a given patient is likely to have persistent functional impairment. The potential clinical utility of fMRI is influenced by and must be weighed against cost and accessibility factors. Future evaluations of the utility of fMRI in prognostic and treatment response studies may consider including a health economics analysis.

A primer on the use of machine learning to distil knowledge from data in biological psychiatry

Applications of machine learning in the biomedical sciences are growing rapidly. This growth has been spurred by diverse cross-institutional and interdisciplinary collaborations, public availability of large datasets, an increase in the accessibility of analytic routines, and the availability of powerful computing resources. With this increased access and exposure to machine learning comes a responsibility for education and a deeper understanding of its bases and bounds, borne equally by data scientists seeking to ply their analytic wares in medical research and by biomedical scientists seeking to harness such methods to glean knowledge from data. This article provides an accessible and critical review of machine learning for a biomedically informed audience, as well as its applications in psychiatry. The review covers definitions and expositions of commonly used machine learning methods, and historical trends of their use in psychiatry. We also provide a set of standards, namely Guidelines for REporting Machine Learning Investigations in Neuropsychiatry (GREMLIN), for designing and reporting studies that use machine learning as a primary data-analysis approach. Lastly, we propose the establishment of the Machine Learning in Psychiatry (MLPsych) Consortium, enumerate its objectives, and identify areas of opportunity for future applications of machine learning in biological psychiatry. This review serves as a cautiously optimistic primer on machine learning for those on the precipice as they prepare to dive into the field, either as methodological practitioners or well-informed consumers.

Predictive power of gait and gait-related cognitive measures in amnestic mild cognitive impairment: a machine learning analysis

Introduction

Gait disorders and gait-related cognitive tests were recently linked to future Alzheimer's Disease (AD) dementia diagnosis in amnestic Mild Cognitive Impairment (aMCI). This study aimed to evaluate the predictive power of gait disorders and gait-related neuropsychological performances for future AD diagnosis in aMCI through machine learning (ML).

Methods

A sample of 253 aMCI (stable, converter) individuals were included. We explored the predictive accuracy of four predictors (gait profile plus MMSE, DSST, and TMT-B) previously identified as critical for the conversion from aMCI to AD within a 36-month follow-up. Supervised ML algorithms (Support Vector Machine [SVM], Logistic Regression, and k-Nearest Neighbors) were trained on 70% of the dataset, and feature importance was evaluated for the best algorithm.

Results

The SVM algorithm achieved the best performance. The optimized training set performance achieved an accuracy of 0.67 (sensitivity = 0.72; specificity = 0.60), improving to 0.70 on the test set (sensitivity = 0.79; specificity = 0.52). Feature importance revealed MMSE as the most important predictor in both training and testing, while gait type was important in the testing phase.

Discussion

We created a predictive ML model that is capable of identifying aMCI at high risk of AD dementia within 36 months. Our ML model could be used to quickly identify individuals at higher risk of AD, facilitating secondary prevention (e.g., cognitive and/or physical training), and serving as screening for more expansive and invasive tests. Lastly, our results point toward theoretically and practically sound evidence of mind and body interaction in AD.

Ant colony optimization for parallel test assembly

Ant colony optimization (ACO) algorithms have previously been used to compile single short scales of psychological constructs. In the present article, we showcase the versatility of the ACO to construct multiple parallel short scales that adhere to several competing and interacting criteria simultaneously. Based on an initial pool of 120 knowledge items, we assembled three 12-item tests that (a) adequately cover the construct at the domain level, (b) follow a unidimensional measurement model, (c) allow reliable and (d) precise measurement of factual knowledge, and (e) are gender-fair. Moreover, we aligned the test characteristic and test information functions of the three tests to establish the equivalence of the tests. We cross-validated the assembled short scales and investigated their association with the full scale and covariates that were not included in the optimization procedure. Finally, we discuss potential extensions to metaheuristic test assembly and the equivalence of parallel knowledge tests in general.

A multivariate cognitive approach to predict social functioning in recent onset psychosis in response to computerized cognitive training

Clinical and neuroimaging data has been increasingly used in recent years to disentangle heterogeneity of treatment response to cognitive training (CT) and predict which individuals may achieve the highest benefits. CT has small to medium effects on improving cognitive and social functioning in recent onset psychosis (ROP) patients, who show the most profound cognitive and social functioning deficits among psychiatric patients. We employed multivariate pattern analysis (MVPA) to investigate the potential of cognitive data to predict social functioning improvement in response to 10 h of CT in patients with ROP. A support vector machine (SVM) classifier was trained on the naturalistic data of the Personalized Prognostic Tools for Early Psychosis Management (PRONIA) study sample to predict functioning in an independent sample of 70 ROP patients using baseline cognitive data. PRONIA is a part of a FP7 EU grant program that involved 7 sites across 5 European countries, designed and conducted with the main aim of identifying (bio)markers associated with an enhanced risk of developing psychosis in order to improve early detection and prognosis. Social functioning was predicted with a balanced accuracy (BAC) of 66.4% (Sensitivity 78.8%; Specificity 54.1%; PPV 60.5%; NPV 74.1%; AUC 0.64; P = 0.01). The most frequently selected cognitive features (mean feature weights > ± 0.2) included the (1) correct number of symbol matchings within the Digit Symbol Substitution Test, (2) the number of distracting stimuli leading to an error within 300 and 200 trials in the Continuous Performance Test and (3) the dynamics of verbal fluency between 15 and 30 s within the Verbal Fluency Test, phonetic part. Next, the SVM classifier generated on the PRONIA sample was applied to the intervention sample, that obtained 54 ROP patients who were randomly assigned to a social cognitive training (SCT) or treatment as usual (TAU) group and dichotomized into good (GF-S ≥ 7) and poor (GF-S < 7) functioning patients based on their level of Global Functioning-Social (GF-S) score at follow-up (FU). By applying the initial PRONIA classifier, using out-of-sample cross-validation (OOCV) to the sample of ROP patients who have undergone the CT intervention, a BAC of 59.3% (Sensitivity 70.4%; Specificity 48.1%; PPV 57.6%; NPV 61.9%; AUC 0.63) was achieved at T0 and a BAC of 64.8% (Sensitivity 66.7%; Specificity 63.0%; PPV 64.3%; NPV 65.4%; AUC 0.66) at FU. After SCT intervention, a significant improvement in predicted social functioning values was observed in the SCT compared to TAU group (P ≤0.05; ES[Cohens' d] = 0.18). Due to a small sample size and modest variance of social functioning of the intervention sample it was not feasible to predict individual response to SCT in the current study. Our findings suggest that the use of baseline cognitive data could provide a robust individual estimate of future social functioning, while prediction of individual response to SCT using cognitive data that can be generated in the routine patient care remains to be addressed in large-scale cognitive training trials.

Translating the Transdiagnostic: Aligning Assessment Practices With Research Advances

Researchers and clinicians working within the Diagnostic and Statistical Manual of Mental Disorders: Fifth Edition, Text Rev (DSM-5-TR) framework face a difficult question: what does it mean to have an evidence-based assessment of a nonevidence-based diagnostic construct? Alternative nosological approaches conceptualize psychopathology as (a) hierarchical, allowing researchers to move between levels of description and (b) dimensional, eliminating artificial dichotomies between disorders and the dichotomy between mental illness and mental well-being. In this article, we provide an overview of ongoing efforts to develop validated measures of transdiagnostic nosologies (i.e., the Hierarchical Taxonomy of Psychopathology; HiTOP) with applications for measurement-based care. However, descriptive models like HiTOP, which summarize patterns of covariation among psychopathology symptoms, do not address dynamic processes underlying the problems associated with psychopathology. Ambulatory assessment, well-suited to examine such dynamic processes, has also developed rapidly in recent decades. Thus, the goal of the current article is twofold. First, we provide a brief overview of developments in constructing valid measures of the HiTOP model as well as developments in ambulatory assessment practices. Second, we outline how these parallel developments can be integrated to advance measurement-based treatment. We end with a discussion of some major challenges for future research to address to integrate advances more fully in transdiagnostic and ambulatory assessment practices.

Prediction of estimated risk for bipolar disorder using machine learning and structural MRI features

BACKGROUND

Individuals with bipolar disorder are commonly correctly diagnosed a decade after symptom onset. Machine learning techniques may aid in early recognition and reduce the disease burden. As both individuals at risk and those with a manifest disease display structural brain markers, structural magnetic resonance imaging may provide relevant classification features.

METHODS

Following a pre-registered protocol, we trained linear support vector machine (SVM) to classify individuals according to their estimated risk for bipolar disorder using regional cortical thickness of help-seeking individuals from seven study sites (N = 276). We estimated the risk using three state-of-the-art assessment instruments (BPSS-P, BARS, EPIbipolar).

RESULTS

For BPSS-P, SVM achieved a fair performance of Cohen's κ of 0.235 (95% CI 0.11-0.361) and a balanced accuracy of 63.1% (95% CI 55.9-70.3) in the 10-fold cross-validation. In the leave-one-site-out cross-validation, the model performed with a Cohen's κ of 0.128 (95% CI -0.069 to 0.325) and a balanced accuracy of 56.2% (95% CI 44.6-67.8). BARS and EPIbipolar could not be predicted. In post hoc analyses, regional surface area, subcortical volumes as well as hyperparameter optimization did not improve the performance.

CONCLUSIONS

Individuals at risk for bipolar disorder, as assessed by BPSS-P, display brain structural alterations that can be detected using machine learning. The achieved performance is comparable to previous studies which attempted to classify patients with manifest disease and healthy controls. Unlike previous studies of bipolar risk, our multicenter design permitted a leave-one-site-out cross-validation. Whole-brain cortical thickness seems to be superior to other structural brain features.

CERAD-NAB and flexible battery based neuropsychological differentiation of Alzheimer's dementia and depression using machine learning approaches

Depression (DEP) and dementia of the Alzheimer's type (DAT) represent the most common neuropsychiatric disorders in elderly patients. Accurate differential diagnosis is indispensable to ensure appropriate treatment. However, DEP can yet mimic cognitive symptoms of DAT and patients with DAT often also present with depressive symptoms, impeding correct diagnosis. Machine learning (ML) approaches could eventually improve this discrimination using neuropsychological test data, but evidence is still missing. We therefore employed Support Vector Machine (SVM), Naïve Bayes (NB), Random Forest (RF) and conventional Logistic Regression (LR) to retrospectively predict the diagnoses of 189 elderly patients (68 DEP and 121 DAT) based on either the well-established Consortium to Establish a Registry for Alzheimer's Disease neuropsychological assessment battery (CERAD-NAB) or a flexible battery approach (FLEXBAT). The best performing combination consisted of FLEXBAT and NB, correctly classifying 87.0% of patients as either DAT or DEP. However, all accuracies were similar across algorithms and test batteries (83.0% - 87.0%). Accordingly, our study is the first to show that common ML algorithms with their default parameters can accurately differentiate between patients clinically diagnosed with DAT or DEP using neuropsychological test data, but do not necessarily outperform conventional LR.

Prediction of antidepressant responses to non-invasive brain stimulation using frontal electroencephalogram signals: Cross-dataset comparisons and validation

BACKGROUND

10-Hz repetitive transcranial magnetic stimulation(rTMS) and intermittent theta-burst stimulation(iTBS) over left prefrontal cortex are FDA-approved, effective options for treatment-resistant depression (TRD). Optimal prediction models for iTBS and rTMS remain elusive. Therefore, our primary objective was to compare prediction accuracy between classification by frontal theta activity alone and machine learning(ML) models by linear and non-linear frontal signals. The second objective was to study an optimal ML model for predicting responses to rTMS and iTBS.

METHODS

Two rTMS and iTBS datasets (n = 163) were used: one randomized controlled trial dataset (RCTD; n = 96) and one outpatient dataset (OPD; n = 67). Frontal theta and non-linear EEG features that reflect trend, stability, and complexity were extracted. Pretreatment frontal EEG and ML algorithms, including classical support vector machine(SVM), random forest(RF), XGBoost, and CatBoost, were analyzed. Responses were defined as ≥50 % depression improvement after treatment. Response rates between those with and without pretreatment prediction in another independent outpatient cohort (n = 208) were compared.

RESULTS

Prediction accuracy using combined EEG features by SVM was better than frontal theta by logistic regression. The accuracy for OPD patients significantly dropped using the RCTD-trained SVM model. Modern ML models, especially RF (rTMS = 83.3 %, iTBS = 88.9 %, p-value(ACC > NIR) < 0.05 for iTBS), performed significantly above chance and had higher accuracy than SVM using both selected features (p < 0.05, FDR corrected for multiple comparisons) or all EEG features. Response rates among those receiving prediction before treatment were significantly higher than those without prediction (p = 0.035).

CONCLUSION

The first study combining linear and non-linear EEG features could accurately predict responses to left PFC iTBS. The bootstraps-based ML model (i.e., RF) had the best predictive accuracy for rTMS and iTBS.

The Adoption of AI in Mental Health Care-Perspectives From Mental Health Professionals: Qualitative Descriptive Study

BACKGROUND

Artificial intelligence (AI) is transforming the mental health care environment. AI tools are increasingly accessed by clients and service users. Mental health professionals must be prepared not only to use AI but also to have conversations about it when delivering care. Despite the potential for AI to enable more efficient and reliable and higher-quality care delivery, there is a persistent gap among mental health professionals in the adoption of AI.

OBJECTIVE

A needs assessment was conducted among mental health professionals to (1) understand the learning needs of the workforce and their attitudes toward AI and (2) inform the development of AI education curricula and knowledge translation products.

METHODS

A qualitative descriptive approach was taken to explore the needs of mental health professionals regarding their adoption of AI through semistructured interviews. To reach maximum variation sampling, mental health professionals (eg, psychiatrists, mental health nurses, educators, scientists, and social workers) in various settings across Ontario (eg, urban and rural, public and private sector, and clinical and research) were recruited.

RESULTS

A total of 20 individuals were recruited. Participants included practitioners (9/20, 45% social workers and 1/20, 5% mental health nurses), educator scientists (5/20, 25% with dual roles as professors/lecturers and researchers), and practitioner scientists (3/20, 15% with dual roles as researchers and psychiatrists and 2/20, 10% with dual roles as researchers and mental health nurses). Four major themes emerged: (1) fostering practice change and building self-efficacy to integrate AI into patient care; (2) promoting system-level change to accelerate the adoption of AI in mental health; (3) addressing the importance of organizational readiness as a catalyst for AI adoption; and (4) ensuring that mental health professionals have the education, knowledge, and skills to harness AI in optimizing patient care.

CONCLUSIONS

AI technologies are starting to emerge in mental health care. Although many digital tools, web-based services, and mobile apps are designed using AI algorithms, mental health professionals have generally been slower in the adoption of AI. As indicated by this study's findings, the implications are 3-fold. At the individual level, digital professionals must see the value in digitally compassionate tools that retain a humanistic approach to care. For mental health professionals, resistance toward AI adoption must be acknowledged through educational initiatives to raise awareness about the relevance, practicality, and benefits of AI. At the organizational level, digital professionals and leaders must collaborate on governance and funding structures to promote employee buy-in. At the societal level, digital and mental health professionals should collaborate in the creation of formal AI training programs specific to mental health to address knowledge gaps. This study promotes the design of relevant and sustainable education programs to support the adoption of AI within the mental health care sphere.

Power Analysis for the Wald, LR, Score, and Gradient Tests in a Marginal Maximum Likelihood Framework: Applications in IRT

The Wald, likelihood ratio, score, and the recently proposed gradient statistics can be used to assess a broad range of hypotheses in item response theory models, for instance, to check the overall model fit or to detect differential item functioning. We introduce new methods for power analysis and sample size planning that can be applied when marginal maximum likelihood estimation is used. This allows the application to a variety of IRT models, which are commonly used in practice, e.g., in large-scale educational assessments. An analytical method utilizes the asymptotic distributions of the statistics under alternative hypotheses. We also provide a sampling-based approach for applications where the analytical approach is computationally infeasible. This can be the case with 20 or more items, since the computational load increases exponentially with the number of items. We performed extensive simulation studies in three practically relevant settings, i.e., testing a Rasch model against a 2PL model, testing for differential item functioning, and testing a partial credit model against a generalized partial credit model. The observed distributions of the test statistics and the power of the tests agreed well with the predictions by the proposed methods in sufficiently large samples. We provide an openly accessible R package that implements the methods for user-supplied hypotheses.

Using machine learning to predict sudden gains in intensive treatment for PTSD

Sudden gains have been found in PTSD treatment across samples and treatment modality. Sudden gains have consistently predicted better treatment response, illustrating clear clinical implications, though attempts to identify predictors of sudden gains have produced inconsistent findings. To date, sudden gains have not been examined in intensive PTSD treatment programs (ITPs). This study explored the occurrence of sudden gains in a 3-week and 2-week ITP (n = 465 and n = 235), evaluated the effect of sudden gains on post-treatment and follow-up PTSD severity while controlling for overall change, and used three machine learning algorithms to assess our ability to predict sudden gains. We found 31% and 19% of our respective samples experienced a sudden gain during the ITP. In both ITPs, sudden gain status predicted greater PTSD symptom improvement at post-treatment (t2 W=-8.57, t3 W=-14.86, p < .001) and at 3-month follow-up (t2 W=-3.82, t3 W=-5.32, p < .001). However, the effect for follow-up was no longer significant after controlling for total symptom reduction across the ITP (t2 W=-1.59, t3 W=-0.32, p > .05). Our ability to predict sudden gains was poor (AUC <.7) across all three machine learning algorithms. These findings demonstrate that sudden gains can be detected in intensive treatment for PTSD, though their implications for treatment outcomes may be limited. Moreover, despite the use of three machine-learning methods across two fairly large clinical samples, we were still unable to identify variables that accurately predict whether an individual will experience a sudden gain during treatment. Implications for clinical application of these findings and for future studies are discussed.

Opening the Black Box of Family-Based Treatments: An Artificial Intelligence Framework to Examine Therapeutic Alliance and Therapist Empathy

The evidence-based treatment (EBT) movement has primarily focused on core intervention content or treatment fidelity and has largely ignored practitioner skills to manage interpersonal process issues that emerge during treatment, especially with difficult-to-treat adolescents (delinquent, substance-using, medical non-adherence) and those of color. A chief complaint of "real world" practitioners about manualized treatments is the lack of correspondence between following a manual and managing microsocial interpersonal processes (e.g. negative affect) that arise in treating "real world clients." Although family-based EBTs share core similarities (e.g. focus on family interactions, emphasis on practitioner engagement, family involvement), most of these treatments do not have an evidence base regarding common implementation and treatment process problems that practitioners experience in delivering particular models, especially in mid-treatment when demands on families to change their behavior is greatest in treatment - a lack that characterizes the field as a whole. Failure to effectively address common interpersonal processes with difficult-to-treat families likely undermines treatment fidelity and sustained use of EBTs, treatment outcome, and contributes to treatment dropout and treatment nonadherence. Recent advancements in wearables, sensing technologies, multivariate time-series analyses, and machine learning allow scientists to make significant advancements in the study of psychotherapy processes by looking "under the skin" of the provider-client interpersonal interactions that define therapeutic alliance, empathy, and empathic accuracy, along with the predictive validity of these therapy processes (therapeutic alliance, therapist empathy) to treatment outcome. Moreover, assessment of these processes can be extended to develop procedures for training providers to manage difficult interpersonal processes while maintaining a physiological profile that is consistent with astute skills in psychotherapeutic processes. This paper argues for opening the "black box" of therapy to advance the science of evidence-based psychotherapy by examining the clinical interior of evidence-based treatments to develop the next generation of audit- and feedback- (i.e., systemic review of professional performance) supervision systems.

Effects of Substance Use and Antisocial Personality on Neuroimaging-Based Machine Learning Prediction of Schizophrenia

BACKGROUND AND HYPOTHESIS

Neuroimaging-based machine learning (ML) algorithms have the potential to aid the clinical diagnosis of schizophrenia. However, literature on the effect of prevalent comorbidities such as substance use disorder (SUD) and antisocial personality (ASPD) on these models' performance has remained unexplored. We investigated whether the presence of SUD or ASPD affects the performance of neuroimaging-based ML models trained to discern patients with schizophrenia (SCH) from controls.

STUDY DESIGN

We trained an ML model on structural MRI data from public datasets to distinguish between SCH and controls (SCH = 347, controls = 341). We then investigated the model's performance in two independent samples of individuals undergoing forensic psychiatric examination: sample 1 was used for sensitivity analysis to discern ASPD (N = 52) from SCH (N = 66), and sample 2 was used for specificity analysis to discern ASPD (N = 26) from controls (N = 25). Both samples included individuals with SUD.

STUDY RESULTS

In sample 1, 94.4% of SCH with comorbid ASPD and SUD were classified as SCH, followed by patients with SCH + SUD (78.8% classified as SCH) and patients with SCH (60.0% classified as SCH). The model failed to discern SCH without comorbidities from ASPD + SUD (AUC = 0.562, 95%CI = 0.400-0.723). In sample 2, the model's specificity to predict controls was 84.0%. In both samples, about half of the ASPD + SUD were misclassified as SCH. Data-driven functional characterization revealed associations between the classification as SCH and cognition-related brain regions.

CONCLUSION

Altogether, ASPD and SUD appear to have effects on ML prediction performance, which potentially results from converging cognition-related brain abnormalities between SCH, ASPD, and SUD.