The promise of machine learning in predicting treatment outcomes in psychiatry

Electronic health records and stratified psychiatry: bridge to precision treatment?

The use of a stratified psychiatry approach that combines electronic health records (EHR) data with machine learning (ML) is one potentially fruitful path toward rapidly improving precision treatment in clinical practice. This strategy, however, requires confronting pervasive methodological flaws as well as deficiencies in transparency and reporting in the current conduct of ML-based studies for treatment prediction. EHR data shares many of the same data quality issues as other types of data used in ML prediction, plus some unique challenges. To fully leverage EHR data's power for patient stratification, increased attention to data quality and collection of patient-reported outcome data is needed.

The Psychedelic Future of Post-Traumatic Stress Disorder Treatment

Post-traumatic stress disorder (PTSD) is a mental health condition that can occur following exposure to a traumatic experience. An estimated 12 million U.S. adults are presently affected by this disorder. Current treatments include psychological therapies (e.g., exposure-based interventions) and pharmacological treatments (e.g., selective serotonin reuptake inhibitors (SSRIs)). However, a significant proportion of patients receiving standard-of-care therapies for PTSD remain symptomatic, and new approaches for this and other trauma-related mental health conditions are greatly needed. Psychedelic compounds that alter cognition, perception, and mood are currently being examined for their efficacy in treating PTSD despite their current status as Drug Enforcement Administration (DEA)- scheduled substances. Initial clinical trials have demonstrated the potential value of psychedelicassisted therapy to treat PTSD and other psychiatric disorders. In this comprehensive review, we summarize the state of the science of PTSD clinical care, including current treatments and their shortcomings. We review clinical studies of psychedelic interventions to treat PTSD, trauma-related disorders, and common comorbidities. The classic psychedelics psilocybin, lysergic acid diethylamide (LSD), and N,N-dimethyltryptamine (DMT) and DMT-containing ayahuasca, as well as the entactogen 3,4-methylenedioxymethamphetamine (MDMA) and the dissociative anesthetic ketamine, are reviewed. For each drug, we present the history of use, psychological and somatic effects, pharmacology, and safety profile. The rationale and proposed mechanisms for use in treating PTSD and traumarelated disorders are discussed. This review concludes with an in-depth consideration of future directions for the psychiatric applications of psychedelics to maximize therapeutic benefit and minimize risk in individuals and communities impacted by trauma-related conditions.

Digitalising mental health care: Practical recommendations from the European Psychiatric Association

The digitalisation of mental health care is expected to improve the accessibility and quality of specialised treatment services and introduce innovative methods to study, assess, and monitor mental health disorders. In this narrative review and practical recommendation of the European Psychiatric Association (EPA), we aim to help healthcare providers and policymakers to navigate this rapidly evolving field. We provide an overview of the current scientific and implementation status across two major domains of digitalisation: i) digital mental health interventions and ii) digital phenotyping, discuss the potential of each domain to improve the accessibility and outcomes of mental health services, and highlight current challenges faced by researchers, clinicians, and service users. Furthermore, we make several recommendations meant to foster the widespread adoption of evidence-based digital solutions for mental health care in the member states of the EPA. To realise the vision of a digitalised, patient-centred, and data-driven mental health ecosystem, a number of implementation challenges must be considered and addressed, spanning from human, technical, ethical-legal, and economic barriers. The list of priority areas and action points our expert panel has identified could serve as a playbook for this process.

Finding the Best Match - a Case Study on the (Text-)Feature and Model Choice in Digital Mental Health Interventions

With the need for psychological help long exceeding the supply, finding ways of scaling, and better allocating mental health support is a necessity. This paper contributes by investigating how to best predict intervention dropout and failure to allow for a need-based adaptation of treatment. We systematically compare the predictive power of different text representation methods (metadata, TF-IDF, sentiment and topic analysis, and word embeddings) in combination with supplementary numerical inputs (socio-demographic, evaluation, and closed-question data). Additionally, we address the research gap of which ML model types - ranging from linear to sophisticated deep learning models - are best suited for different features and outcome variables. To this end, we analyze nearly 16.000 open-text answers from 849 German-speaking users in a Digital Mental Health Intervention (DMHI) for stress. Our research proves that - contrary to previous findings - there is great promise in using neural network approaches on DMHI text data. We propose a task-specific LSTM-based model architecture to tackle the challenge of long input sequences and thereby demonstrate the potential of word embeddings (AUC scores of up to 0.7) for predictions in DMHIs. Despite the relatively small data set, sequential deep learning models, on average, outperform simpler features such as metadata and bag-of-words approaches when predicting dropout. The conclusion is that user-generated text of the first two sessions carries predictive power regarding patients' dropout and intervention failure risk. Furthermore, the match between the sophistication of features and models needs to be closely considered to optimize results, and additional non-text features increase prediction results.

Supplementary Information

The online version contains supplementary material available at 10.1007/s41666-023-00148-z.

Deep learning for the prediction of clinical outcomes in internet-delivered CBT for depression and anxiety

In treating depression and anxiety, just over half of all clients respond. Monitoring and obtaining early client feedback can allow for rapidly adapted treatment delivery and improve outcomes. This study seeks to develop a state-of-the-art deep-learning framework for predicting clinical outcomes in internet-delivered Cognitive Behavioural Therapy (iCBT) by leveraging large-scale, high-dimensional time-series data of client-reported mental health symptoms and platform interaction data. We use de-identified data from 45,876 clients on SilverCloud Health, a digital platform for the psychological treatment of depression and anxiety. We train deep recurrent neural network (RNN) models to predict whether a client will show reliable improvement by the end of treatment using clinical measures, interaction data with the iCBT program, or both. Outcomes are based on total improvement in symptoms of depression (Patient Health Questionnaire-9, PHQ-9) and anxiety (Generalized Anxiety Disorder-7, GAD-7), as reported within the iCBT program. Using internal and external datasets, we compare the proposed models against several benchmarks and rigorously evaluate them according to their predictive accuracy, sensitivity, specificity and AUROC over treatment. Our proposed RNN models consistently predict reliable improvement in PHQ-9 and GAD-7, using past clinical measures alone, with above 87% accuracy and 0.89 AUROC after three or more review periods, outperforming all benchmark models. Additional evaluations demonstrate the robustness of the achieved models across (i) different health services; (ii) geographic locations; (iii) iCBT programs, and (iv) client severity subgroups. Results demonstrate the robust performance of dynamic prediction models that can yield clinically helpful prognostic information ready for implementation within iCBT systems to support timely decision-making and treatment adjustments by iCBT clinical supporters towards improved client outcomes.

[Personalized psychotherapy of posttraumatic stress disorder : Overview on the selection of treatment methods and techniques using statistical procedures]

BACKGROUND

A relevant heterogeneity of treatment effects in posttraumatic stress disorder (PTSD) is discussed with respect to the debate about the necessity of phase-based treatment and in light of the new diagnosis of complex PTSD and has recently been proven; however, there has been little personalization in the treatment of PTSD. This article presents the current state of research on the personalized selection of specific psychotherapeutic methods for the treatment of PTSD based on patient characteristics using statistical methods.

METHODS

A systematic literature search was conducted in the PubMed (including Medline), Embase, Web of Science Core Collection, Google Scholar, PsycINFO and PSYNDEX databases to identify clinical trials and reviews examining personalized treatment for PTSD.

RESULTS

A total of 13 relevant publications were identified, of which 5 articles were predictor analyses in samples without control conditions and 7 articles showed analyses of randomized controlled trials (RCT) with a post hoc comparison of treatment effects in optimally and nonoptimally assigned patients. In addition, one article was a systematic review on the treatment of patients with comorbid borderline personality order and PTSD.

DISCUSSION

The available manuscripts indicate the importance and benefits of personalized treatment in PTSD. The relevant predictor variables identified for personalization should be used as a suggestion to investigate them in future prospective studies.

Using machine learning to forecast symptom changes among subclinical depression patients receiving stepped care or usual care

BACKGROUND

Subclinical depression (SD) is a mental health disorder characterized by minor depressive symptoms. Most SD patients are treated in the primary practice, but many respond poorly to treatment at the expense of provider resources. Stepped care approaches are appealing for tiering SD care to efficiently allocate scarce resources while jointly optimizing patient outcomes. However, stepped care can be time inefficient, as some persons may respond poorly and be forced to suffer with their symptoms for prolonged periods. Machine learning can offer insight into optimal treatment paths and inform clinical recommendations for incident patients.

METHODS

As part of the Step-Dep trial, participants with SD were randomized to receive stepped care (N=96) or usual care (N=140). Machine learning was used to predict changes in depressive symptoms every three months over a year for each treatment group.

RESULTS

Tree-based models were effective in predicting PHQ-9 changes among patients who received stepped care (r=0.35-0.46, MAE=0.14-0.17) and usual care (r=0.34-0.49, MAE=0.15-0.18). Patients who received stepped care were more likely to reduce PHQ-9 scores if they had high PHQ-9 but low HADS-A scores at baseline, a low number of chronic illnesses, and an internal locus of control.

LIMITATIONS

Models may suffer from potential overfitting due to sample size limitations.

CONCLUSION

Our findings demonstrate the promise of machine learning for predicting changes in depressive symptoms for SD patients receiving different treatments. Trained models can intake incident patient information and predict outcomes to inform personalized care.

Next-generation cognitive assessment: Combining functional brain imaging, system perturbations and novel equipment interfaces

Conventional cognitive assessment is widely used in clinical and research settings, in educational institutions, and in the corporate world for personnel selection. Such approaches involve having a client, a patient, or a research participant complete a series of standardized cognitive tasks in order to challenge specific and global cognitive abilities, and then quantify performance for the desired end purpose. The latter may include a diagnostic confirmation of a disease, description of a state or ability, or matching cognitive characteristics to a particular occupational role requirement. Metrics derived from cognitive assessments are putatively informative about important features of the brain and its function. For this reason, the research sector also makes use of cognitive assessments, most frequently as a stimulus for cognitive activity from which to extract functional neuroimaging data. Such "task-related activations" form the core of the most widely used neuroimaging technologies, such as fMRI. Much of what we know about the brain has been drawn from the interleaving of cognitive assessments of various types with functional brain imaging technologies. Despite innovation in neuroimaging (i.e., quantifying the neural response), relatively little innovation has occurred on task presentation and volitional response measurement; yet these together comprise the core of cognitive performance. Moreover, even when cognitive assessment is interleaved with functional neuroimaging, this is most often undertaken in the research domain, rather than the primary applications of cognitive assessment in diagnosis and monitoring, education and personnel selection. There are new ways in which brain imaging-and even more importantly, brain modulation-technologies can be combined with automation and artificial intelligence to deliver next-generation cognitive assessment methods. In this review paper, we describe some prototypes for how this can be done and identify important areas for progress (technological and otherwise) to enable it to happen. We will argue that the future of cognitive assessment will include semi- and fully-automated assessments involving neuroimaging, standardized perturbations via neuromodulation technologies, and artificial intelligence. Furthermore, the fact that cognitive assessments take place in a social/interpersonal context-normally between the patient and clinician-makes the human-machine interface consequential, and this will also be discussed.

Challenges and Ethical Considerations to Successfully Implement Artificial Intelligence in Clinical Medicine and Neuroscience: a Narrative Review

This narrative review discusses how the safe and effective use of clinical artificial intelligence (AI) prediction tools requires recognition of the importance of human intelligence. Human intelligence, creativity, situational awareness, and professional knowledge, are required for successful implementation. The implementation of clinical AI prediction tools may change the workflow in medical practice resulting in new challenges and safety implications. Human understanding of how a clinical AI prediction tool performs in routine and exceptional situations is fundamental to successful implementation. Physicians must be involved in all aspects of the selection, implementation, and ongoing product monitoring of clinical AI prediction tools.

Digital Phenotyping: Data-Driven Psychiatry to Redefine Mental Health

The term "digital phenotype" refers to the digital footprint left by patient-environment interactions. It has potential for both research and clinical applications but challenges our conception of health care by opposing 2 distinct approaches to medicine: one centered on illness with the aim of classifying and curing disease, and the other centered on patients, their personal distress, and their lived experiences. In the context of mental health and psychiatry, the potential benefits of digital phenotyping include creating new avenues for treatment and enabling patients to take control of their own well-being. However, this comes at the cost of sacrificing the fundamental human element of psychotherapy, which is crucial to addressing patients' distress. In this viewpoint paper, we discuss the advances rendered possible by digital phenotyping and highlight the risk that this technology may pose by partially excluding health care professionals from the diagnosis and therapeutic process, thereby foregoing an essential dimension of care. We conclude by setting out concrete recommendations on how to improve current digital phenotyping technology so that it can be harnessed to redefine mental health by empowering patients without alienating them.

Exercise as an add-on treatment in individuals with schizophrenia: Results from a large multicenter randomized controlled trial

Current treatment methods do not achieve recovery for most individuals with schizophrenia, and symptoms such as negative symptoms and cognitive deficits often persist. Aerobic endurance training has been suggested as a potential add-on treatment targeting both physical and mental health. We performed a large-scale multicenter, rater-blind, parallel-group randomized controlled clinical trial in individuals with stable schizophrenia. Participants underwent a professionally supervised six-month training comprising either aerobic endurance training (AET) or flexibility, strengthening, and balance training (FSBT, control group), follow-up was another six months. The primary endpoint was all-cause discontinuation (ACD); secondary endpoints included effects on psychopathology, cognition, functioning, and cardiovascular risk. In total, 180 participants were randomized. AET was not superior to FSBT in ACD and most secondary outcomes, with dropout rates of 59.55% and 57.14% in the six-month active phase, respectively. However, both groups showed significant improvements in positive, general, and total symptoms, levels of functioning and in cognitive performance. A higher training frequency additionally promoted further memory domains. Participants with higher baseline cognitive abilities were more likely to respond to the interventions. Our results support integrating exercise into schizophrenia treatment, while future studies should aim to develop personalized training recommendations to maximize exercise-induced benefits.

Treatment-resistant depression: definition, prevalence, detection, management, and investigational interventions

Treatment-resistant depression (TRD) is common and associated with multiple serious public health implications. A consensus definition of TRD with demonstrated predictive utility in terms of clinical decision-making and health outcomes does not currently exist. Instead, a plethora of definitions have been proposed, which vary significantly in their conceptual framework. The absence of a consensus definition hampers precise estimates of the prevalence of TRD, and also belies efforts to identify risk factors, prevention opportunities, and effective interventions. In addition, it results in heterogeneity in clinical practice decision-making, adversely affecting quality of care. The US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) have adopted the most used definition of TRD (i.e., inadequate response to a minimum of two antidepressants despite adequacy of the treatment trial and adherence to treatment). It is currently estimated that at least 30% of persons with depression meet this definition. A significant percentage of persons with TRD are actually pseudo-resistant (e.g., due to inadequacy of treatment trials or non-adherence to treatment). Although multiple sociodemographic, clinical, treatment and contextual factors are known to negatively moderate response in persons with depression, very few factors are regarded as predictive of non-response across multiple modalities of treatment. Intravenous ketamine and intranasal esketamine (co-administered with an antidepressant) are established as efficacious in the management of TRD. Some second-generation antipsychotics (e.g., aripiprazole, brexpiprazole, cariprazine, quetiapine XR) are proven effective as adjunctive treatments to antidepressants in partial responders, but only the olanzapine-fluoxetine combination has been studied in FDA-defined TRD. Repetitive transcranial magnetic stimulation (TMS) is established as effective and FDA-approved for individuals with TRD, with accelerated theta-burst TMS also recently showing efficacy. Electroconvulsive therapy is regarded as an effective acute and maintenance intervention in TRD, with preliminary evidence suggesting non-inferiority to acute intravenous ketamine. Evidence for extending antidepressant trial, medication switching and combining antidepressants is mixed. Manual-based psychotherapies are not established as efficacious on their own in TRD, but offer significant symptomatic relief when added to conventional antidepressants. Digital therapeutics are under study and represent a potential future clinical vista in this population.

Machine learning-based prediction of delirium 24 h after pediatric intensive care unit admission in critically ill children: A prospective cohort study

BACKGROUND

Accurately identifying patients at high risk of delirium is vital for timely preventive intervention measures. Approaches for identifying the risk of developing delirium among critically ill children are not well researched.

OBJECTIVE

To develop and validate machine learning-based models for predicting delirium among critically ill children 24 h after pediatric intensive care unit (PICU) admission.

DESIGN

A prospective cohort study.

SETTING

A large academic medical center with a 57-bed PICU in southwestern China from November 2019 to February 2022.

PARTICIPANTS

One thousand five hundred and seventy-six critically ill children requiring PICU stay over 24 h.

METHODS

Five machine learning algorithms were employed. Delirium was screened by bedside nurses twice a day using the Cornell Assessment of Pediatric Delirium. Twenty-four clinical features from medical and nursing records during hospitalization were used to inform the models. Model performance was assessed according to numerous learning metrics, including the area under the receiver operating characteristic curve (AUC).

RESULTS

Of the 1576 enrolled patients, 929 (58.9 %) were boys, and the age ranged from 28 days to 15 years with a median age of 12 months (IQR 3 to 60 months). Among them, 1126 patients were assigned to the training cohort, and 450 were assigned to the validation cohort. The AUCs ranged from 0.763 to 0.805 for the five models, among which the eXtreme Gradient Boosting (XGB) model performed best, achieving an AUC of 0.805 (95 % CI, 0.759-0.851), with 0.798 (95 % CI, 0.758-0.834) accuracy, 0.902 sensitivity, 0.839 positive predictive value, 0.640 F1-score and a Brier score of 0.144. Almost all models showed lower predictive performance in children younger than 24 months than in older children. The logistic regression model also performed well, with an AUC of 0.789 (95 % CI, 0.739, 0.838), just under that of the XGB model, and was subsequently transformed into a nomogram.

CONCLUSIONS

Machine learning-based models can be established and potentially help identify critically ill children who are at high risk of delirium 24 h after PICU admission. The nomogram may be a beneficial management tool for delirium for PICU practitioners at present.

Harnessing Artificial Intelligence: Strategies for Mental Health Nurses in Optimizing Psychiatric Patient Care

This narrative review explores the transformative impact of Artificial Intelligence (AI) on mental health nursing, particularly in enhancing psychiatric patient care. AI technologies present new strategies for early detection, risk assessment, and improving treatment adherence in mental health. They also facilitate remote patient monitoring, bridge geographical gaps, and support clinical decision-making. The evolution of virtual mental health assistants and AI-enhanced therapeutic interventions are also discussed. These technological advancements reshape the nurse-patient interactions while ensuring personalized, efficient, and high-quality care. The review also addresses AI's ethical and responsible use in mental health nursing, emphasizing patient privacy, data security, and the balance between human interaction and AI tools. As AI applications in mental health care continue to evolve, this review encourages continued innovation while advocating for responsible implementation, thereby optimally leveraging the potential of AI in mental health nursing.

Machine learning prediction will be part of future treatment of depression

Machine learning (ML) is changing the way that medicine is practiced. While already clinically utilised in diagnostic radiology and outcome prediction in intensive care unit, ML approaches in psychiatry remain nascent. Implementing ML algorithms in psychiatry, particularly in the treatment of depression, is significantly more challenging than other areas of medicine in part because of the less demarcated disease nosology and greater variability in practice. Given the current exiguous capacity of clinicians to predict patient and treatment outcomes in depression, there is a significantly greater need for better predictive capability. Early studies have shown promising results. ML predictions were significantly better than chance within the sequenced treatment alternatives to relieve depression (STAR*D) trial (accuracy 64.6%, p < 0.0001) and combining medications to enhance depression outcomes (COMED) randomised Controlled Trial (RCT) (accuracy 59.6%, p = 0.043), with similar results found in larger scale, retrospective studies. The greater flexibility and dimensionality of ML approaches has been demonstrated in studies incorporating diverse input variables including electroencephalography scans, achieving 88% accuracy for treatment response, and cognitive test scores, achieving up to 72% accuracy for treatment response. The predicting response to depression treatment (PReDicT) trial tested ML informed prescribing of antidepressants against standard therapy and found there was both better outcomes for anxiety and functional endpoints despite the algorithm only having a balanced accuracy of 57.5%. Impeding the progress of ML algorithms in psychiatry are pragmatic hurdles, including accuracy, expense, acceptability and comprehensibility, and ethical hurdles, including medicolegal liability, clinical autonomy and data privacy. Notwithstanding impediments, it is clear that ML prediction algorithms will be part of depression treatment in the future and clinicians should be prepared for their arrival.

Modulation of resting-state functional connectivity in default mode network is associated with the long-term treatment outcome in major depressive disorder

BACKGROUND

Treatment non-response and recurrence are the main sources of disease burden in major depressive disorder (MDD). However, little is known about its neurobiological mechanism concerning the brain network changes accompanying pharmacotherapy. The present study investigated the changes in the intrinsic brain networks during 6-month antidepressant treatment phase associated with the treatment response and recurrence in MDD.

METHODS

Resting-state functional magnetic resonance imaging was acquired from untreated patients with MDD and healthy controls at baseline. The patients' depressive symptoms were monitored by using the Hamilton Rating Scale for Depression (HAMD). After 6 months of antidepressant treatment, patients were re-scanned and followed up every 6 months over 2 years. Traditional statistical analysis as well as machine learning approaches were conducted to investigate the longitudinal changes in macro-scale resting-state functional network connectivity (rsFNC) strength and micro-scale resting-state functional connectivity (rsFC) associated with long-term treatment outcome in MDD.

RESULTS

Repeated measures of the general linear model demonstrated a significant difference in the default mode network (DMN) rsFNC change before and after the 6-month antidepressant treatment between remitters and non-remitters. The difference in the rsFNC change over the 6-month antidepressant treatment between recurring and stable MDD was also specific to DMN. Machine learning analysis results revealed that only the DMN rsFC change successfully distinguished non-remitters from the remitters at 6 months and recurring from stable MDD during the 2-year follow-up.

CONCLUSION

Our findings demonstrated that the intrinsic DMN connectivity could be a unique and important target for treatment and recurrence prevention in MDD.

Poor response to methylphenidate is associated with a smaller dorsal attentive network in adult Attention-Deficit/Hyperactivity Disorder (ADHD)

Stimulants, such as methylphenidate (MPH), are effective in treating attention-deficit/hyperactivity disorder (ADHD), but there is individual variability in response, especially in adults. To improve outcomes, we need to understand the factors associated with adult treatment response. This longitudinal study investigated whether pre-treatment anatomy of the fronto-striatal and fronto-parietal attentional networks was associated with MPH treatment response. 60 adults with ADHD underwent diffusion brain imaging before starting MPH treatment, and response was measured at two months. We tested the association between brain anatomy and treatment response by using regression-based approaches; and compared the identified anatomical characteristics with those of 20 matched neurotypical controls in secondary analyses. Finally, we explored whether combining anatomical with clinical and neuropsychological data through machine learning provided a more comprehensive profile of factors associated with treatment response. At a group level, a smaller left dorsal superior longitudinal fasciculus (SLF I), a tract responsible for the voluntary control of attention, was associated with a significantly lower probability of being responders to two-month MPH-treatment. The association between the volume of the left SLF I and treatment response was driven by improvement on both inattentive and hyperactive/impulsive symptoms. Only non-responders significantly differed from controls in this tract metric. Finally, our machine learning approach identified clinico-neuropsychological factors associated with treatment response, such as higher cognitive performance and symptom severity at baseline. These novel findings add to our understanding of the pathophysiological mechanisms underlying response to MPH, pointing to the dorsal attentive network as playing a key role.

Duration of Untreated Illness in Patients with Obsessive-Compulsive Disorder and Its Impact on Long-Term Outcome: A Systematic Review

Background: Duration of untreated illness (DUI)-defined as the time period between the onset of a mental disorder and its first adequate treatment-should influence patients' long-term prognosis and outcome. In patients with obsessive-compulsive disorder (OCD), DUI lasts on average from 87.5 up to 94.5 months, being significantly longer compared with data available from patients affected by other severe mental disorders, such as schizophrenia and bipolar disorder. We carried out a systematic review in order to assess the impact of DUI on long-term outcomes in OCD patients. Methods: A systemic review has been implemented, searching from inception to April 2023; only papers written in English were included. Results: Seventy-one articles were initially identified; only eight papers were included in the review. The DUI ranged from 7.0 ± 8.5 to 20.9 ± 11.2 years. Patients reporting a longer DUI have a poor long-term outcome in terms of lower level of treatment response and greater symptom severity. Conclusions: The present review confirms that longer DUI has a negative impact on the long-term outcome of patients with OCD. It should be useful to promote the dissemination of early interventions with a specific focus on OCD symptoms.

Revolutionizing stress-related disorder regulation through neuroinformatics and data analysis: An editorial

Collapse

Closing the Gap for Children with OCD: A Staged-Care Model of Cognitive Behavioural Therapy with Exposure and Response Prevention

Childhood obsessive-compulsive disorder (OCD) is among the most prevalent and disabling mental health conditions affecting children and adolescents. Although the distress and burden associated with childhood OCD are well documented and empirically supported treatments are available, there remains an unacceptable "treatment gap" and "quality gap" in the provision of services for youth suffering from OCD. The treatment gap represents the large number of children who never receive mental health services for OCD, while the quality gap refers to the children and young people who do access services, but do not receive evidence-based, cognitive behavioural therapy with exposure and response prevention (CBT-ERP). We propose a novel staged-care model of CBT-ERP that aims to improve the treatment access to high-quality CBT-ERP, as well as enhance the treatment outcomes for youth. In staged care, patients receive hierarchically arranged service packages that vary according to the intensity, duration, and mix of treatment options, with provision of care from prevention, early intervention, through to first and second-line treatments. Based on a comprehensive review of the literature on treatment outcomes and predictors of treatments response, we propose a preliminary staging algorithm to determine the level of clinical care, informed by three key determinants: severity of illness, comorbidity, and prior treatment history. The proposed clinical staging model for paediatric OCD prioritises high-quality care for children at all stages and levels of illness, utilising empirically supported CBT-ERP, across multiple modalities, combined with evidence-informed, clinical decision-making heuristics. While informed by evidence, the proposed staging model requires empirical validation before it is ready for prime time.

Moving toward precision PTSD treatment: predicting veterans' intensive PTSD treatment response using continuously updating machine learning models

BACKGROUND

Considerable heterogeneity exists in treatment response to first-line posttraumatic stress disorder (PTSD) treatments, such as Cognitive Processing Therapy (CPT). Relatively little is known about the timing of when during a course of care the treatment response becomes apparent. Novel machine learning methods, especially continuously updating prediction models, have the potential to address these gaps in our understanding of response and optimize PTSD treatment.

METHODS

Using data from a 3-week (n = 362) CPT-based intensive PTSD treatment program (ITP), we explored three methods for generating continuously updating prediction models to predict endpoint PTSD severity. These included Mixed Effects Bayesian Additive Regression Trees (MixedBART), Mixed Effects Random Forest (MERF) machine learning models, and Linear Mixed Effects models (LMM). Models used baseline and self-reported PTSD symptom severity data collected every other day during treatment. We then validated our findings by examining model performances in a separate, equally established, 2-week CPT-based ITP (n = 108).

RESULTS

Results across approaches were very similar and indicated modest prediction accuracy at baseline (R2 ~ 0.18), with increasing accuracy of predictions of final PTSD severity across program timepoints (e.g. mid-program R2 ~ 0.62). Similar findings were obtained when the models were applied to the 2-week ITP. Neither the MERF nor the MixedBART machine learning approach outperformed LMM prediction, though benefits of each may differ based on the application.

CONCLUSIONS

Utilizing continuously updating models in PTSD treatments may be beneficial for clinicians in determining whether an individual is responding, and when this determination can be made.

Prediction of depression treatment outcome from multimodal data: a CAN-BIND-1 report

BACKGROUND

Prediction of treatment outcomes is a key step in improving the treatment of major depressive disorder (MDD). The Canadian Biomarker Integration Network in Depression (CAN-BIND) aims to predict antidepressant treatment outcomes through analyses of clinical assessment, neuroimaging, and blood biomarkers.

METHODS

In the CAN-BIND-1 dataset of 192 adults with MDD and outcomes of treatment with escitalopram, we applied machine learning models in a nested cross-validation framework. Across 210 analyses, we examined combinations of predictive variables from three modalities, measured at baseline and after 2 weeks of treatment, and five machine learning methods with and without feature selection. To optimize the predictors-to-observations ratio, we followed a tiered approach with 134 and 1152 variables in tier 1 and tier 2 respectively.

RESULTS

A combination of baseline tier 1 clinical, neuroimaging, and molecular variables predicted response with a mean balanced accuracy of 0.57 (best model mean 0.62) compared to 0.54 (best model mean 0.61) in single modality models. Adding week 2 predictors improved the prediction of response to a mean balanced accuracy of 0.59 (best model mean 0.66). Adding tier 2 features did not improve prediction.

CONCLUSIONS

A combination of clinical, neuroimaging, and molecular data improves the prediction of treatment outcomes over single modality measurement. The addition of measurements from the early stages of treatment adds precision. Present results are limited by lack of external validation. To achieve clinically meaningful prediction, the multimodal measurement should be scaled up to larger samples and the robustness of prediction tested in an external validation dataset.

Out of their minds? Externalist challenges for using AI in forensic psychiatry

Harnessing the power of machine learning (ML) and other Artificial Intelligence (AI) techniques promises substantial improvements across forensic psychiatry, supposedly offering more objective evaluations and predictions. However, AI-based predictions about future violent behaviour and criminal recidivism pose ethical challenges that require careful deliberation due to their social and legal significance. In this paper, we shed light on these challenges by considering externalist accounts of psychiatric disorders which stress that the presentation and development of psychiatric disorders is intricately entangled with their outward environment and social circumstances. We argue that any use of predictive AI in forensic psychiatry should not be limited to neurobiology alone but must also consider social and environmental factors. This thesis has practical implications for the design of predictive AI systems, especially regarding the collection and processing of training data, the selection of ML methods, and the determination of their explainability requirements.

Major depressive disorder

Major depressive disorder (MDD) is characterized by persistent depressed mood, loss of interest or pleasure in previously enjoyable activities, recurrent thoughts of death, and physical and cognitive symptoms. People with MDD can have reduced quality of life owing to the disorder itself as well as related medical comorbidities, social factors, and impaired functional outcomes. MDD is a complex disorder that cannot be fully explained by any one single established biological or environmental pathway. Instead, MDD seems to be caused by a combination of genetic, environmental, psychological and biological factors. Treatment for MDD commonly involves pharmacological therapy with antidepressant medications, psychotherapy or a combination of both. In people with severe and/or treatment-resistant MDD, other biological therapies, such as electroconvulsive therapy, may also be offered.

Digital transformation of mental health services

This paper makes a case for digital mental health and provides insights into how digital technologies can enhance (but not replace) existing mental health services. We describe digital mental health by presenting a suite of digital technologies (from digital interventions to the application of artificial intelligence). We discuss the benefits of digital mental health, for example, a digital intervention can be an accessible stepping-stone to receiving support. The paper does, however, present less-discussed benefits with new concepts such as 'poly-digital', where many different apps/features (e.g. a sleep app, mood logging app and a mindfulness app, etc.) can each address different factors of wellbeing, perhaps resulting in an aggregation of marginal gains. Another benefit is that digital mental health offers the ability to collect high-resolution real-world client data and provide client monitoring outside of therapy sessions. These data can be collected using digital phenotyping and ecological momentary assessment techniques (i.e. repeated mood or scale measures via an app). This allows digital mental health tools and real-world data to inform therapists and enrich face-to-face sessions. This can be referred to as blended care/adjunctive therapy where service users can engage in 'channel switching' between digital and non-digital (face-to-face) interventions providing a more integrated service. This digital integration can be referred to as a kind of 'digital glue' that helps join up the in-person sessions with the real world. The paper presents the challenges, for example, the majority of mental health apps are maybe of inadequate quality and there is a lack of user retention. There are also ethical challenges, for example, with the perceived 'over-promotion' of screen-time and the perceived reduction in care when replacing humans with 'computers', and the trap of 'technological solutionism' whereby technology can be naively presumed to solve all problems. Finally, we argue for the need to take an evidence-based, systems thinking and co-production approach in the form of stakeholder-centred design when developing digital mental health services based on technologies. The main contribution of this paper is the integration of ideas from many different disciplines as well as the framework for blended care using 'channel switching' to showcase how digital data and technology can enrich physical services. Another contribution is the emergence of 'poly-digital' and a discussion on the challenges of digital mental health, specifically 'digital ethics'.

Predictors of treatment dropout in patients with posttraumatic stress disorder due to childhood abuse

Background

Knowledge about patient characteristics predicting treatment dropout for post-traumatic stress disorder (PTSD) is scarce, whereas more understanding about this topic may give direction to address this important issue.

Method

Data were obtained from a randomized controlled trial in which a phase-based treatment condition (Eye Movement Desensitization and Reprocessing [EMDR] therapy preceded by Skills Training in Affect and Interpersonal Regulation [STAIR]; n = 57) was compared with a direct trauma-focused treatment (EMDR therapy only; n = 64) in people with a PTSD due to childhood abuse. All pre-treatment variables included in the trial were examined as possible predictors for dropout using machine learning techniques.

Results

For the dropout prediction, a model was developed using Elastic Net Regularization. The ENR model correctly predicted dropout in 81.6% of all individuals. Males, with a low education level, suicidal thoughts, problems in emotion regulation, high levels of general psychopathology and not using benzodiazepine medication at screening proved to have higher scores on dropout.

Conclusion

Our results provide directions for the development of future programs in addition to PTSD treatment or for the adaptation of current treatments, aiming to reduce treatment dropout among patients with PTSD due to childhood abuse.

Improving mental health care in depression: A call for action

Depressive disorders have one of the highest disability-adjusted life years (DALYs) of all medical conditions, which led the European Psychiatric Association to propose a policy paper, pinpointing their unmet health care and research needs. The first part focuses on what can be currently done to improve the care of patients with depression, and then discuss future trends for research and healthcare. Through the narration of clinical cases, the different points are illustrated. The necessary political framework is formulated, to implement such changes to fundamentally improve psychiatric care. The group of European Psychiatrist Association (EPA) experts insist on the need for (1) increased awareness of mental illness in primary care settings, (2) the development of novel (biological) markers, (3) the rapid implementation of machine learning (supporting diagnostics, prognostics, and therapeutics), (4) the generalized use of electronic devices and apps into everyday treatment, (5) the development of the new generation of treatment options, such as plasticity-promoting agents, and (6) the importance of comprehensive recovery approach. At a political level, the group also proposed four priorities, the need to (1) increase the use of open science, (2) implement reasonable data protection laws, (3) establish ethical electronic health records, and (4) enable better healthcare research and saving resources.

Development of a model to predict antidepressant treatment response for depression among Veterans

BACKGROUND

Only a limited number of patients with major depressive disorder (MDD) respond to a first course of antidepressant medication (ADM). We investigated the feasibility of creating a baseline model to determine which of these would be among patients beginning ADM treatment in the US Veterans Health Administration (VHA).

METHODS

A 2018-2020 national sample of n = 660 VHA patients receiving ADM treatment for MDD completed an extensive baseline self-report assessment near the beginning of treatment and a 3-month self-report follow-up assessment. Using baseline self-report data along with administrative and geospatial data, an ensemble machine learning method was used to develop a model for 3-month treatment response defined by the Quick Inventory of Depression Symptomatology Self-Report and a modified Sheehan Disability Scale. The model was developed in a 70% training sample and tested in the remaining 30% test sample.

RESULTS

In total, 35.7% of patients responded to treatment. The prediction model had an area under the ROC curve (s.e.) of 0.66 (0.04) in the test sample. A strong gradient in probability (s.e.) of treatment response was found across three subsamples of the test sample using training sample thresholds for high [45.6% (5.5)], intermediate [34.5% (7.6)], and low [11.1% (4.9)] probabilities of response. Baseline symptom severity, comorbidity, treatment characteristics (expectations, history, and aspects of current treatment), and protective/resilience factors were the most important predictors.

CONCLUSIONS

Although these results are promising, parallel models to predict response to alternative treatments based on data collected before initiating treatment would be needed for such models to help guide treatment selection.

Attitudes Toward the Adoption of 2 Artificial Intelligence-Enabled Mental Health Tools Among Prospective Psychotherapists: Cross-sectional Study

BACKGROUND

Despite growing efforts to develop user-friendly artificial intelligence (AI) applications for clinical care, their adoption remains limited because of the barriers at individual, organizational, and system levels. There is limited research on the intention to use AI systems in mental health care.

OBJECTIVE

This study aimed to address this gap by examining the predictors of psychology students' and early practitioners' intention to use 2 specific AI-enabled mental health tools based on the Unified Theory of Acceptance and Use of Technology.

METHODS

This cross-sectional study included 206 psychology students and psychotherapists in training to examine the predictors of their intention to use 2 AI-enabled mental health care tools. The first tool provides feedback to the psychotherapist on their adherence to motivational interviewing techniques. The second tool uses patient voice samples to derive mood scores that the therapists may use for treatment decisions. Participants were presented with graphic depictions of the tools' functioning mechanisms before measuring the variables of the extended Unified Theory of Acceptance and Use of Technology. In total, 2 structural equation models (1 for each tool) were specified, which included direct and mediated paths for predicting tool use intentions.

RESULTS

Perceived usefulness and social influence had a positive effect on the intention to use the feedback tool (P<.001) and the treatment recommendation tool (perceived usefulness, P=.01 and social influence, P<.001). However, trust was unrelated to use intentions for both the tools. Moreover, perceived ease of use was unrelated (feedback tool) and even negatively related (treatment recommendation tool) to use intentions when considering all predictors (P=.004). In addition, a positive relationship between cognitive technology readiness (P=.02) and the intention to use the feedback tool and a negative relationship between AI anxiety and the intention to use the feedback tool (P=.001) and the treatment recommendation tool (P<.001) were observed.

CONCLUSIONS

The results shed light on the general and tool-dependent drivers of AI technology adoption in mental health care. Future research may explore the technological and user group characteristics that influence the adoption of AI-enabled tools in mental health care.

A framework for precision "dosing" of mental healthcare services: algorithm development and clinical pilot

BACKGROUND

One in five adults in the US experience mental illness and over half of these adults do not receive treatment. In addition to the access gap, few innovations have been reported for ensuring the right level of mental healthcare service is available at the right time for individual patients.

METHODS

Historical observational clinical data was leveraged from a virtual healthcare system. We conceptualize mental healthcare services themselves as therapeutic interventions and develop a prototype computational framework to estimate their potential longitudinal impacts on depressive symptom severity, which is then used to assess new treatment schedules and delivered to clinicians via a dashboard. We operationally define this process as "session dosing": 497 patients who started treatment with severe symptoms of depression between November 2020 and October 2021 were used for modeling. Subsequently, 22 mental health providers participated in a 5-week clinical quality improvement (QI) pilot, where they utilized the prototype dashboard in treatment planning with 126 patients.

RESULTS

The developed framework was able to resolve patient symptom fluctuations from their treatment schedules: 77% of the modeling dataset fit criteria for using the individual fits for subsequent clinical planning where five anecdotal profile types were identified that presented different clinical opportunities. Based on initial quality thresholds for model fits, 88% of those individuals were identified as adequate for session optimization planning using the developed dashboard, while 12% supported more thorough treatment planning (e.g. different treatment modalities). In the clinical pilot, 90% of clinicians reported using the dashboard a few times or more per member. Although most clinicians (67.5%) either rarely or never used the dashboard to change session types, numerous other discussions were enabled, and opportunities for automating session recommendations were identified.

CONCLUSIONS

It is possible to model and identify the extent to which mental healthcare services can resolve depressive symptom severity fluctuations. Implementation of one such prototype framework in a real-world clinic represents an advancement in mental healthcare treatment planning; however, investigations to assess which clinical endpoints are impacted by this technology, and the best way to incorporate such frameworks into clinical workflows, are needed and are actively being pursued.

Sampling inequalities affect generalization of neuroimaging-based diagnostic classifiers in psychiatry

BACKGROUND

The development of machine learning models for aiding in the diagnosis of mental disorder is recognized as a significant breakthrough in the field of psychiatry. However, clinical practice of such models remains a challenge, with poor generalizability being a major limitation.

METHODS

Here, we conducted a pre-registered meta-research assessment on neuroimaging-based models in the psychiatric literature, quantitatively examining global and regional sampling issues over recent decades, from a view that has been relatively underexplored. A total of 476 studies (n = 118,137) were included in the current assessment. Based on these findings, we built a comprehensive 5-star rating system to quantitatively evaluate the quality of existing machine learning models for psychiatric diagnoses.

RESULTS

A global sampling inequality in these models was revealed quantitatively (sampling Gini coefficient (G) = 0.81, p < .01), varying across different countries (regions) (e.g., China, G = 0.47; the USA, G = 0.58; Germany, G = 0.78; the UK, G = 0.87). Furthermore, the severity of this sampling inequality was significantly predicted by national economic levels (β =  - 2.75, p < .001, R²_adj = 0.40; r =  - .84, 95% CI: - .41 to - .97), and was plausibly predictable for model performance, with higher sampling inequality for reporting higher classification accuracy. Further analyses showed that lack of independent testing (84.24% of models, 95% CI: 81.0-87.5%), improper cross-validation (51.68% of models, 95% CI: 47.2-56.2%), and poor technical transparency (87.8% of models, 95% CI: 84.9-90.8%)/availability (80.88% of models, 95% CI: 77.3-84.4%) are prevailing in current diagnostic classifiers despite improvements over time. Relating to these observations, model performances were found decreased in studies with independent cross-country sampling validations (all p < .001, BF₁₀ > 15). In light of this, we proposed a purpose-built quantitative assessment checklist, which demonstrated that the overall ratings of these models increased by publication year but were negatively associated with model performance.

CONCLUSIONS

Together, improving sampling economic equality and hence the quality of machine learning models may be a crucial facet to plausibly translating neuroimaging-based diagnostic classifiers into clinical practice.

The role of pharmacogenomics in precision psychiatry

The field of psychiatry is facing an important paradigm shift in the provision of clinical care and mental health service organization toward personalization and integration of multimodal data science. This approach, termed precision psychiatry, aims at identifying subgroups of patients more prone to the development of a certain phenotype, such as symptoms or severe mental disorders (risk detection), and/or to guide treatment selection. Pharmacogenomics and computational psychiatry are two fundamental tools of precision psychiatry, which have seen increasing levels of integration in clinical settings. Here we present a brief overview of these two applications of precision psychiatry in clinical settings.

Multivariate resting-state functional connectomes predict and characterize obesity phenotypes

The univariate obesity-brain associations have been extensively explored, while little is known about the multivariate associations between obesity and resting-state functional connectivity. We therefore utilized machine learning and resting-state functional connectivity to develop and validate predictive models of 4 obesity phenotypes (i.e. body fat percentage, body mass index, waist circumference, and waist-height ratio) in 3 large neuroimaging datasets (n = 2,992). Preliminary evidence suggested that the resting-state functional connectomes effectively predicted obesity/weight status defined by each obesity phenotype with good generalizability to longitudinal and independent datasets. However, the differences between resting-state functional connectivity patterns characterizing different obesity phenotypes indicated that the obesity-brain associations varied according to the type of measure of obesity. The shared structure among resting-state functional connectivity patterns revealed reproducible neuroimaging biomarkers of obesity, primarily comprising the connectomes within the visual cortex and between the visual cortex and inferior parietal lobule, visual cortex and orbital gyrus, and amygdala and orbital gyrus, which further suggested that the dysfunctions in the perception, attention and value encoding of visual information (e.g. visual food cues) and abnormalities in the reward circuit may act as crucial neurobiological bases of obesity. The recruitment of multiple obesity phenotypes is indispensable in future studies seeking reproducible obesity-brain associations.

Predicting new-onset post-stroke depression from real-world data using machine learning algorithm

Introduction

Post-stroke depression (PSD) is a serious mental disorder after ischemic stroke. Early detection is important for clinical practice. This research aims to develop machine learning models to predict new-onset PSD using real-world data.

Methods

We collected data for ischemic stroke patients from multiple medical institutions in Taiwan between 2001 and 2019. We developed models from 61,460 patients and used 15,366 independent patients to test the models' performance by evaluating their specificities and sensitivities. The predicted targets were whether PSD occurred at 30, 90, 180, and 365 days post-stroke. We ranked the important clinical features in these models.

Results

In the study's database sample, 1.3% of patients were diagnosed with PSD. The average specificity and sensitivity of these four models were 0.83-0.91 and 0.30-0.48, respectively. Ten features were listed as important features related to PSD at different time points, namely old age, high height, low weight post-stroke, higher diastolic blood pressure after stroke, no pre-stroke hypertension but post-stroke hypertension (new-onset hypertension), post-stroke sleep-wake disorders, post-stroke anxiety disorders, post-stroke hemiplegia, and lower blood urea nitrogen during stroke.

Discussion

Machine learning models can provide as potential predictive tools for PSD and important factors are identified to alert clinicians for early detection of depression in high-risk stroke patients.

Candidate biomarkers in psychiatric disorders: state of the field

The field of psychiatry is hampered by a lack of robust, reliable and valid biomarkers that can aid in objectively diagnosing patients and providing individualized treatment recommendations. Here we review and critically evaluate the evidence for the most promising biomarkers in the psychiatric neuroscience literature for autism spectrum disorder, schizophrenia, anxiety disorders and post-traumatic stress disorder, major depression and bipolar disorder, and substance use disorders. Candidate biomarkers reviewed include various neuroimaging, genetic, molecular and peripheral assays, for the purposes of determining susceptibility or presence of illness, and predicting treatment response or safety. This review highlights a critical gap in the biomarker validation process. An enormous societal investment over the past 50 years has identified numerous candidate biomarkers. However, to date, the overwhelming majority of these measures have not been proven sufficiently reliable, valid and useful to be adopted clinically. It is time to consider whether strategic investments might break this impasse, focusing on a limited number of promising candidates to advance through a process of definitive testing for a specific indication. Some promising candidates for definitive testing include the N170 signal, an event-related brain potential measured using electroencephalography, for subgroup identification within autism spectrum disorder; striatal resting-state functional magnetic resonance imaging (fMRI) measures, such as the striatal connectivity index (SCI) and the functional striatal abnormalities (FSA) index, for prediction of treatment response in schizophrenia; error-related negativity (ERN), an electrophysiological index, for prediction of first onset of generalized anxiety disorder, and resting-state and structural brain connectomic measures for prediction of treatment response in social anxiety disorder. Alternate forms of classification may be useful for conceptualizing and testing potential biomarkers. Collaborative efforts allowing the inclusion of biosystems beyond genetics and neuroimaging are needed, and online remote acquisition of selected measures in a naturalistic setting using mobile health tools may significantly advance the field. Setting specific benchmarks for well-defined target application, along with development of appropriate funding and partnership mechanisms, would also be crucial. Finally, it should never be forgotten that, for a biomarker to be actionable, it will need to be clinically predictive at the individual level and viable in clinical settings.

Waiting for a digital therapist: three challenges on the path to psychotherapy delivered by artificial intelligence

Growing demand for broadly accessible mental health care, together with the rapid development of new technologies, trigger discussions about the feasibility of psychotherapeutic interventions based on interactions with Conversational Artificial Intelligence (CAI). Many authors argue that while currently available CAI can be a useful supplement for human-delivered psychotherapy, it is not yet capable of delivering fully fledged psychotherapy on its own. The goal of this paper is to investigate what are the most important obstacles on our way to developing CAI systems capable of delivering psychotherapy in the future. To this end, we formulate and discuss three challenges central to this quest. Firstly, we might not be able to develop effective AI-based psychotherapy unless we deepen our understanding of what makes human-delivered psychotherapy effective. Secondly, assuming that it requires building a therapeutic relationship, it is not clear whether psychotherapy can be delivered by non-human agents. Thirdly, conducting psychotherapy might be a problem too complicated for narrow AI, i.e., AI proficient in dealing with only relatively simple and well-delineated tasks. If this is the case, we should not expect CAI to be capable of delivering fully-fledged psychotherapy until the so-called "general" or "human-like" AI is developed. While we believe that all these challenges can ultimately be overcome, we think that being mindful of them is crucial to ensure well-balanced and steady progress on our path to AI-based psychotherapy.

Machine Learning Techniques for Developing Remotely Monitored Central Nervous System Biomarkers Using Wearable Sensors: A Narrative Literature Review

BACKGROUND

Central nervous system (CNS) disorders benefit from ongoing monitoring to assess disease progression and treatment efficacy. Mobile health (mHealth) technologies offer a means for the remote and continuous symptom monitoring of patients. Machine Learning (ML) techniques can process and engineer mHealth data into a precise and multidimensional biomarker of disease activity.

OBJECTIVE

This narrative literature review aims to provide an overview of the current landscape of biomarker development using mHealth technologies and ML. Additionally, it proposes recommendations to ensure the accuracy, reliability, and interpretability of these biomarkers.

METHODS

This review extracted relevant publications from databases such as PubMed, IEEE, and CTTI. The ML methods employed across the selected publications were then extracted, aggregated, and reviewed.

RESULTS

This review synthesized and presented the diverse approaches of 66 publications that address creating mHealth-based biomarkers using ML. The reviewed publications provide a foundation for effective biomarker development and offer recommendations for creating representative, reproducible, and interpretable biomarkers for future clinical trials.

CONCLUSION

mHealth-based and ML-derived biomarkers have great potential for the remote monitoring of CNS disorders. However, further research and standardization of study designs are needed to advance this field. With continued innovation, mHealth-based biomarkers hold promise for improving the monitoring of CNS disorders.

Systematic review of machine learning utilization within outpatient psychodynamic psychotherapy research

Introduction

Although outpatient psychodynamic psychotherapy is effective, there has been no improvement in treatment success in recent years. One way to improve psychodynamic treatment could be the use of machine learning to design treatments tailored to the individual patient's needs. In the context of psychotherapy, machine learning refers mainly to various statistical methods, which aim to predict outcomes (e.g., drop-out) of future patients as accurately as possible. We therefore searched various literature for all studies using machine learning in outpatient psychodynamic psychotherapy research to identify current trends and objectives.

Methods

For this systematic review, we applied the Preferred Reporting Items for systematic Reviews and Meta-Analyses Guidelines.

Results

In total, we found four studies that used machine learning in outpatient psychodynamic psychotherapy research. Three of these studies were published between 2019 and 2021.

Discussion

We conclude that machine learning has only recently made its way into outpatient psychodynamic psychotherapy research and researchers might not yet be aware of its possible uses. Therefore, we have listed a variety of perspectives on how machine learning could be used to increase treatment success of psychodynamic psychotherapies. In doing so, we hope to give new impetus to outpatient psychodynamic psychotherapy research on how to use machine learning to address previously unsolved problems.

Subjective and objective sleep and circadian parameters as predictors of depression-related outcomes: A machine learning approach in UK Biobank

BACKGROUND

Sleep and circadian disruption are associated with depression onset and severity, but it is unclear which features (e.g., sleep duration, chronotype) are important and whether they can identify individuals showing poorer outcomes.

METHODS

Within a subset of the UK Biobank with actigraphy and mental health data (n = 64,353), penalised regression identified the most useful of 51 sleep/rest-activity predictors of depression-related outcomes; including case-control (Major Depression (MD) vs. controls; postnatal depression vs. controls) and within-case comparisons (severe vs. moderate MD; early vs. later onset, atypical vs. typical symptoms; comorbid anxiety; suicidality). Best models (of lasso, ridge, and elastic net) were selected based on Area Under the Curve (AUC).

RESULTS

For MD vs. controls (n_(MD) = 24,229; n_(control) = 40,124), lasso AUC was 0.68, 95 % confidence interval (CI) 0.67-0.69. Discrimination was reasonable for atypical vs. typical symptoms (n_(atypical) = 958; n_(typical) = 18,722; ridge: AUC 0.74, 95 % CI 0.71-0.77) but poor for remaining models (AUCs 0.59-0.67). Key predictors across most models included: difficulty getting up, insomnia symptoms, snoring, actigraphy-measured daytime inactivity and lower morning activity (~8 am). In a distinct subset (n = 310,718), the number of these factors shown was associated with all depression outcomes.

LIMITATIONS

Analyses were cross-sectional and in middle-/older aged adults: comparison with longitudinal investigations and younger cohorts is necessary.

DISCUSSION

Sleep and circadian measures alone provided poor to moderate discrimination of depression outcomes, but several characteristics were identified that may be clinically useful. Future work should assess these features alongside broader sociodemographic, lifestyle and genetic features.

Machine learning v. traditional regression models predicting treatment outcomes for binge-eating disorder from a randomized controlled trial

BACKGROUND

While effective treatments exist for binge-eating disorder (BED), prediction of treatment outcomes has proven difficult, and few reliable predictors have been identified. Machine learning is a promising method for improving the accuracy of difficult-to-predict outcomes. We compared the accuracy of traditional and machine-learning approaches for predicting BED treatment outcomes.

METHODS

Participants were 191 adults with BED in a randomized controlled trial testing 6-month behavioral and stepped-care treatments. Outcomes, determined by independent assessors, were binge-eating (% reduction, abstinence), eating-disorder psychopathology, and weight loss (% loss, ⩾5% loss). Predictors included treatment condition, demographic information, and baseline clinical characteristics. Traditional models were logistic/linear regressions. Machine-learning models were elastic net regressions and random forests. Predictive accuracy was indicated by the area under receiver operator characteristic curve (AUC), root mean square error (RMSE), and R2. Confidence intervals were used to compare accuracy across models.

RESULTS

Across outcomes, AUC ranged from very poor to fair (0.49-0.73) for logistic regressions, elastic nets, and random forests, with few significant differences across model types. RMSE was significantly lower for elastic nets and random forests v. linear regressions but R2 values were low (0.01-0.23).

CONCLUSIONS

Different analytic approaches revealed some predictors of key treatment outcomes, but accuracy was limited. Machine-learning models with unbiased resampling methods provided a minimal advantage over traditional models in predictive accuracy for treatment outcomes.

AI-assisted prediction of differential response to antidepressant classes using electronic health records

Antidepressant selection is largely a trial-and-error process. We used electronic health record (EHR) data and artificial intelligence (AI) to predict response to four antidepressants classes (SSRI, SNRI, bupropion, and mirtazapine) 4 to 12 weeks after antidepressant initiation. The final data set comprised 17,556 patients. Predictors were derived from both structured and unstructured EHR data and models accounted for features predictive of treatment selection to minimize confounding by indication. Outcome labels were derived through expert chart review and AI-automated imputation. Regularized generalized linear model (GLM), random forest, gradient boosting machine (GBM), and deep neural network (DNN) models were trained and their performance compared. Predictor importance scores were derived using SHapley Additive exPlanations (SHAP). All models demonstrated similarly good prediction performance (AUROCs ≥ 0.70, AUPRCs ≥ 0.68). The models can estimate differential treatment response probabilities both between patients and between antidepressant classes for the same patient. In addition, patient-specific factors driving response probabilities for each antidepressant class can be generated. We show that antidepressant response can be accurately predicted from real-world EHR data with AI modeling, and our approach could inform further development of clinical decision support systems for more effective treatment selection.

Can Epigenetics Predict Drug Efficiency in Mental Disorders?

Psychiatric disorders affect millions of individuals and their families worldwide, and the costs to society are substantial and are expected to rise due to a lack of effective treatments. Personalized medicine-customized treatment tailored to the individual-offers a solution. Although most mental diseases are influenced by genetic and environmental factors, finding genetic biomarkers that predict treatment efficacy has been challenging. This review highlights the potential of epigenetics as a tool for predicting treatment efficacy and personalizing medicine for psychiatric disorders. We examine previous studies that have attempted to predict treatment efficacy through epigenetics, provide an experimental model, and note the potential challenges at each stage. While the field is still in its infancy, epigenetics holds promise as a predictive tool by examining individual patients' epigenetic profiles in conjunction with other indicators. However, further research is needed, including additional studies, replication, validation, and application beyond clinical settings.

Multimodal predictions of treatment outcome in major depression: A comparison of data-driven predictors with importance ratings by clinicians

BACKGROUND

Reliable prediction models of treatment outcome in Major Depressive Disorder (MDD) are currently lacking in clinical practice. Data-driven outcome definitions, combining data from multiple modalities and incorporating clinician expertise might improve predictions.

METHODS

We used unsupervised machine learning to identify treatment outcome classes in 1060 MDD inpatients. Subsequently, classification models were created on clinical and biological baseline information to predict treatment outcome classes and compared to the performance of two widely used classical outcome definitions. We also related the findings to results from an online survey that assessed which information clinicians use for outcome prognosis.

RESULTS

Three and four outcome classes were identified by unsupervised learning. However, data-driven outcome classes did not result in more accurate prediction models. The best prediction model was targeting treatment response in its standard definition and reached accuracies of 63.9 % in the test sample, and 59.5 % and 56.9 % in the validation samples. Top predictors included sociodemographic and clinical characteristics, while biological parameters did not improve prediction accuracies. Treatment history, personality factors, prior course of the disorder, and patient attitude towards treatment were ranked as most important indicators by clinicians.

LIMITATIONS

Missing data limited the power to identify biological predictors of treatment outcome from certain modalities.

CONCLUSIONS

So far, the inclusion of available biological measures in addition to psychometric and clinical information did not improve predictive value of the models, which was overall low. Optimized biomarkers, stratified predictions and the inclusion of clinical expertise may improve future prediction models.

Individual differences in computational psychiatry: a review of current challenges

Bringing precision to the understanding and treatment of mental disorders requires instruments for studying clinically relevant individual differences. One promising approach is the development of computational assays: integrating computational models with cognitive tasks to infer latent patient-specific disease processes in brain computations. While recent years have seen many methodological advancements in computational modelling and many cross-sectional patient studies, much less attention has been paid to basic psychometric properties (reliability and construct validity) of the computational measures provided by the assays. In this review, we assess the extent of this issue by examining emerging empirical evidence. We find that many computational measures suffer from poor psychometric properties, which poses a risk of invalidating previous findings and undermining ongoing research efforts using computational assays to study individual (and even group) differences. We provide recommendations for how to address these problems and, crucially, embed them within a broader perspective on key developments that are needed for translating computational assays to clinical practice.

Machine learning methods to predict outcomes of pharmacological treatment in psychosis

In recent years, machine learning (ML) has been a promising approach in the research of treatment outcome prediction in psychosis. In this study, we reviewed ML studies using different neuroimaging, neurophysiological, genetic, and clinical features to predict antipsychotic treatment outcomes in patients at different stages of schizophrenia. Literature available on PubMed until March 2022 was reviewed. Overall, 28 studies were included, among them 23 using a single-modality approach and 5 combining data from multiple modalities. The majority of included studies considered structural and functional neuroimaging biomarkers as predictive features used in ML models. Specifically, functional magnetic resonance imaging (fMRI) features contributed to antipsychotic treatment response prediction of psychosis with good accuracies. Additionally, several studies found that ML models based on clinical features might present adequate predictive ability. Importantly, by examining the additive effects of combining features, the predictive value might be improved by applying multimodal ML approaches. However, most of the included studies presented several limitations, such as small sample sizes and a lack of replication tests. Moreover, considerable clinical and analytical heterogeneity among included studies posed a challenge in synthesizing findings and generating robust overall conclusions. Despite the complexity and heterogeneity of methodology, prognostic features, clinical presentation, and treatment approaches, studies included in this review suggest that ML tools may have the potential to predict treatment outcomes of psychosis accurately. Future studies need to focus on refining feature characterization, validating prediction models, and evaluate their translation in real-world clinical practice.

Predicting psychotic relapse following randomised discontinuation of paliperidone in individuals with schizophrenia or schizoaffective disorder: an individual participant data analysis

BACKGROUND

Predicting relapse for individuals with psychotic disorders is not well established, especially after discontinuation of antipsychotic treatment. We aimed to identify general prognostic factors of relapse for all participants (irrespective of treatment continuation or discontinuation) and specific predictors of relapse for treatment discontinuation, using machine learning.

METHODS

For this individual participant data analysis, we searched the Yale University Open Data Access Project's database for placebo-controlled, randomised antipsychotic discontinuation trials with participants with schizophrenia or schizoaffective disorder (aged ≥18 years). We included studies in which participants were treated with any antipsychotic study drug and randomly assigned to continue the same antipsychotic drug or to discontinue it and receive placebo. We assessed 36 prespecified baseline variables at randomisation to predict time to relapse, using univariate and multivariate proportional hazard regression models (including multivariate treatment group by variable interactions) with machine learning to categorise the variables as general prognostic factors of relapse, specific predictors of relapse, or both.

FINDINGS

We identified 414 trials, of which five trials with 700 participants (304 [43%] women and 396 [57%] men) were eligible for the continuation group and 692 participants (292 [42%] women and 400 [58%] men) were eligible for the discontinuation group (median age 37 [IQR 28-47] years for continuation group and 38 [28-47] years for discontinuation group). Out of the 36 baseline variables, general prognostic factors of increased risk of relapse for all participants were drug-positive urine; paranoid, disorganised, and undifferentiated types of schizophrenia (lower risk for schizoaffective disorder); psychiatric and neurological adverse events; higher severity of akathisia (ie, difficulty or inability to sit still); antipsychotic discontinuation; lower social performance; younger age; lower glomerular filtration rate; benzodiazepine comedication (lower risk for anti-epileptic comedication). Out of the 36 baseline variables, predictors of increased risk specifically after antipsychotic discontinuation were increased prolactin concentration, higher number of hospitalisations, and smoking. Both prognostic factors and predictors with increased risk after discontinuation were oral antipsychotic treatment (lower risk for long-acting injectables), higher last dosage of the antipsychotic study drug, shorter duration of antipsychotic treatment, and higher score on the Clinical Global Impression (CGI) severity scale The predictive performance (concordance index) for participants who were not used to train the model was 0·707 (chance level is 0·5).

INTERPRETATION

Routinely available general prognostic factors of psychotic relapse and predictors specific for treatment discontinuation could be used to support personalised treatment. Abrupt discontinuation of higher dosages of oral antipsychotics, especially for individuals with recurring hospitalisations, higher scores on the CGI severity scale, and increased prolactin concentrations, should be avoided to reduce the risk of relapse.

FUNDING

German Research Foundation and Berlin Institute of Health.

Precision Medicine Approaches to Mental Health Care

Developing a more comprehensive understanding of the physiological underpinnings of mental illness, precision medicine has the potential to revolutionize psychiatric care. With recent breakthroughs in next-generation multi-omics technologies and data analytics, it is becoming more feasible to leverage multimodal biomarkers, from genetic variants to neuroimaging biomarkers, to objectify diagnostics and treatment decisions in psychiatry and improve patient outcomes. Ongoing work in precision psychiatry will parallel progress in precision oncology and cardiology to develop an expanded suite of blood- and neuroimaging-based diagnostic tests, empower monitoring of treatment efficacy over time, and reduce patient exposure to ineffective treatments. The emerging model of precision psychiatry has the potential to mitigate some of psychiatry's most pressing issues, including improving disease classification, lengthy treatment duration, and suboptimal treatment outcomes. This narrative-style review summarizes some of the emerging breakthroughs and recurring challenges in the application of precision medicine approaches to mental health care.

Moderators of LGBQ-affirmative cognitive behavioral therapy: ESTEEM is especially effective among Black and Latino sexual minority men

OBJECTIVE

Lesbian, gay, bisexual, and queer (LGBQ)-affirmative cognitive behavioral therapy (CBT) focused on minority stress processes can address gay and bisexual men's transdiagnostic mental and behavioral health concerns. Identifying moderators of treatment outcomes may inform the mechanisms of LGBQ-affirmative CBT and subpopulations who may derive particular benefit.

METHOD

Data were from a clinical trial in which gay and bisexual men with mental and behavioral health concerns were randomized to receive Effective Skills to Empower Effective Men (ESTEEM; an LGBQ-affirmative transdiagnostic CBT; n = 100) or one of two control conditions (n = 154): LGBQ-affirmative community mental health treatment (CMHT) or HIV counseling and testing (HCT). The preregistered outcome was a comorbidity index of depression, anxiety, alcohol/drug problems, and human immunodeficiency virus (HIV) transmission risk behavior at 8-month follow-up (i.e., 4 months postintervention). A two-step exploratory machine learning process was employed for 20 theoretically informed baseline variables identified by study therapists as potential moderators of ESTEEM efficacy. Potential moderators included demographic factors, pretreatment comorbidities, clinical facilitators, and minority stress factors.

RESULTS

Racial/ethnic minority identification, namely as Black or Latino, was the only statistically significant moderator of treatment efficacy (B = -3.23, 95% CI [-5.03, -1.64]), t(197) = -3.88, p < .001. Racially/ethnically minoritized recipients (d = -0.71, p < .001), but not White/non-Latino recipients (d = 0.22, p = .391), had greater reductions in comorbidity index scores in ESTEEM compared to the control conditions. This moderation was driven by improvements in anxiety and alcohol/drug use problems.

DISCUSSION

Black and Latino gay and bisexual men experiencing comorbid mental and behavioral health risks might particularly benefit from a minority stress-focused LGBQ-affirmative CBT. Future research should identify mechanisms for this moderation to inform targeted treatment delivery and dissemination. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Multi-omics data integration methods and their applications in psychiatric disorders

To study mental illness and health, in the past researchers have often broken down their complexity into individual subsystems (e.g., genomics, transcriptomics, proteomics, clinical data) and explored the components independently. Technological advancements and decreasing costs of high throughput sequencing has led to an unprecedented increase in data generation. Furthermore, over the years it has become increasingly clear that these subsystems do not act in isolation but instead interact with each other to drive mental illness and health. Consequently, individual subsystems are now analysed jointly to promote a holistic understanding of the underlying biological complexity of health and disease. Complementing the increasing data availability, current research is geared towards developing novel methods that can efficiently combine the information rich multi-omics data to discover biologically meaningful biomarkers for diagnosis, treatment, and prognosis. However, clinical translation of the research is still challenging. In this review, we summarise conventional and state-of-the-art statistical and machine learning approaches for discovery of biomarker, diagnosis, as well as outcome and treatment response prediction through integrating multi-omics and clinical data. In addition, we describe the role of biological model systems and in silico multi-omics model designs in clinical translation of psychiatric research from bench to bedside. Finally, we discuss the current challenges and explore the application of multi-omics integration in future psychiatric research. The review provides a structured overview and latest updates in the field of multi-omics in psychiatry.

Prediction of Chinese clients' satisfaction with psychotherapy by machine learning

Background

Effective psychotherapy should satisfy the client, but that satisfaction depends on many factors. We do not fully understand the factors that affect client satisfaction with psychotherapy and how these factors synergistically affect a client's psychotherapy experience.

Aims

This study aims to use machine learning to predict Chinese clients' satisfaction with psychotherapy and analyze potential outcome contributors.

Methods

In this cross-sectional investigation, a self-compiled online questionnaire was delivered through the WeChat app. The information of 791 participants who had received psychotherapy was used in the study. A series of features, for example, the participants' demographic features and psychotherapy-related features, were chosen to distinguish between participants satisfied and dissatisfied with the psychotherapy they received. With our dataset, we trained seven supervised machine-learning-based algorithms to implement prediction models.

Results

Among the 791 participants, 619 (78.3%) reported being satisfied with the psychotherapy sessions that they received. The occupation of the clients, the location of psychotherapy, and the form of access to psychotherapy are the three most recognizable features that determined whether clients are satisfied with psychotherapy. The machine-learning model based on the CatBoost achieved the highest prediction performance in classifying satisfied and psychotherapy clients with an F1 score of 0.758.

Conclusion

This study clarified the factors related to clients' satisfaction with psychotherapy, and the machine-learning-based classifier accurately distinguished clients who were satisfied or unsatisfied with psychotherapy. These results will help provide better psychotherapy strategies for specific clients, so they may achieve better therapeutic outcomes.