Evaluating the Machine Learning Literature: A Primer and User's Guide for Psychiatrists

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.

Trajectory Modeling and Response Prediction in Transcranial Magnetic Stimulation for Depression

Repetitive transcranial magnetic stimulation (rTMS) therapy could be improved by better and earlier prediction of response. Latent class mixture (LCMM) and non-linear mixed effects (NLME) modelling have been applied to model the trajectories of antidepressant response (or non-response) to TMS, but it is not known whether such models can predict clinical outcomes. We compared LCMM and NLME approaches to model the antidepressant response to TMS in a naturalistic sample of 238 patients receiving rTMS for treatment resistant depression (TRD), across multiple coils and protocols. We then compared the predictive power of those models. LCMM trajectories were influenced largely by baseline symptom severity, but baseline symptoms provided little predictive power for later antidepressant response. Rather, the optimal LCMM model was a nonlinear two-class model that accounted for baseline symptoms. This model accurately predicted patient response at 4 weeks of treatment (AUC = 0.70, 95% CI = [0.52-0.87]), but not before. NLME offered slightly improved predictive performance at 4 weeks of treatment (AUC = 0.76, 95% CI = [0.58 - 0.94], but likewise, not before. In showing the predictive validity of these approaches to model response trajectories to rTMS, we provided preliminary evidence that trajectory modeling could be used to guide future treatment decisions.

Predicting adolescent suicidal behavior following inpatient discharge using structured and unstructured data

BACKGROUND

The objective was to develop and assess performance of an algorithm predicting suicide-related ICD codes within three months of psychiatric discharge.

METHODS

This prognostic study used a retrospective cohort of EHR data from 2789 youth (12 to 20 years old) hospitalized in a safety net institution in the Northeastern United States. The dataset combined structured data with unstructured data obtained through natural language processing of clinical notes. Machine learning approaches compared gradient boosting to random forest analyses.

RESULTS

Area under the ROC and precision-recall curve were 0.88 and 0.17, respectively, for the final Gradient Boosting model. The cutoff point of the model-generated predicted probabilities of suicide that optimally classified the individual as high risk or not was 0.009. When applying the chosen cutoff (0.009) to the hold-out testing set, the model correctly identified 8 positive cases out of 10, and 418 negative cases out 548. The corresponding performance metrics showed 80 % sensitivity, 76 % specificity, 6 % PPV, 99 % NPV, F-1 score of 0.11, and an accuracy of 76 %.

LIMITATIONS

The data in this study comes from a single health system, possibly introducing bias in the model's algorithm. Thus, the model may have underestimated the incidence of suicidal behavior in the study population. Further research should include multiple system EHRs.

CONCLUSIONS

These performance metrics suggest a benefit to including both unstructured and structured data in design of predictive algorithms for suicidal behavior, which can be integrated into psychiatric services to help assess risk.

Interpretable deep learning model for major depressive disorder assessment based on functional near-infrared spectroscopy

BACKGROUND

Major depressive disorder (MDD) affects a substantial number of individuals worldwide. New approaches are required to improve the diagnosis of MDD, which relies heavily on subjective reports of depression-related symptoms.

AIM

Establish an objective measurement and evaluation of MDD.

METHODS

Functional near-infrared spectroscopy (fNIRS) was used to investigate the brain activity of MDD patients and healthy controls (HCs). Leveraging a sizeable fNIRS dataset of 263 HCs and 251 patients with MDD, including mild to moderate MDD (mMDD; n = 139) and severe MDD (sMDD; n = 77), we developed an interpretable deep learning model for screening MDD and staging its severity.

RESULTS

The proposed deep learning model achieved an accuracy of 80.9% in diagnostic classification and 78.6% in severity staging for MDD. We discerned five channels with the most significant contribution to MDD identification through Shapley additive explanations (SHAP), located in the right medial prefrontal cortex, right dorsolateral prefrontal cortex, right superior temporal gyrus, and left posterior superior frontal cortex. The findings corresponded closely to the features of haemoglobin responses between HCs and individuals with MDD, as we obtained a good discriminative ability for MDD using cortical channels that are related to the disorder, namely the frontal and temporal cortical channels with areas under the curve of 0.78 and 0.81, respectively.

CONCLUSION

Our study demonstrated the potential of integrating the fNIRS system with artificial intelligence algorithms to classify and stage MDD in clinical settings using a large dataset. This approach can potentially enhance MDD assessment and provide insights for clinical diagnosis and intervention.

Closing the loop in psychiatric deep brain stimulation: physiology, psychometrics, and plasticity

Deep brain stimulation (DBS) is an invasive approach to precise modulation of psychiatrically relevant circuits. Although it has impressive results in open-label psychiatric trials, DBS has also struggled to scale to and pass through multi-center randomized trials. This contrasts with Parkinson disease, where DBS is an established therapy treating thousands of patients annually. The core difference between these clinical applications is the difficulty of proving target engagement, and of leveraging the wide range of possible settings (parameters) that can be programmed in a given patient's DBS. In Parkinson's, patients' symptoms change rapidly and visibly when the stimulator is tuned to the correct parameters. In psychiatry, those same changes take days to weeks, limiting a clinician's ability to explore parameter space and identify patient-specific optimal settings. I review new approaches to psychiatric target engagement, with an emphasis on major depressive disorder (MDD). Specifically, I argue that better engagement may come by focusing on the root causes of psychiatric illness: dysfunction in specific, measurable cognitive functions and in the connectivity and synchrony of distributed brain circuits. I overview recent progress in both those domains, and how it may relate to other technologies discussed in companion articles in this issue.

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.

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.

Machine Learning and Pharmacogenomics at the Time of Precision Psychiatry

Traditional medicine and biomedical sciences are reaching a turning point because of the constantly growing impact and volume of Big Data. Machine Learning (ML) techniques and related algorithms play a central role as diagnostic, prognostic, and decision-making tools in this field. Another promising area becoming part of everyday clinical practice is personalized therapy and pharmacogenomics. Applying ML to pharmacogenomics opens new frontiers to tailored therapeutical strategies to help clinicians choose drugs with the best response and fewer side effects, operating with genetic information and combining it with the clinical profile. This systematic review aims to draw up the state-of-the-art ML applied to pharmacogenomics in psychiatry. Our research yielded fourteen papers; most were published in the last three years. The sample comprises 9,180 patients diagnosed with mood disorders, psychoses, or autism spectrum disorders. Prediction of drug response and prediction of side effects are the most frequently considered domains with the supervised ML technique, which first requires training and then testing. The random forest is the most used algorithm; it comprises several decision trees, reduces the training set's overfitting, and makes precise predictions. ML proved effective and reliable, especially when genetic and biodemographic information were integrated into the algorithm. Even though ML and pharmacogenomics are not part of everyday clinical practice yet, they will gain a unique role in the next future in improving personalized treatments in psychiatry.

Modern views of machine learning for precision psychiatry

In light of the National Institute of Mental Health (NIMH)'s Research Domain Criteria (RDoC), the advent of functional neuroimaging, novel technologies and methods provide new opportunities to develop precise and personalized prognosis and diagnosis of mental disorders. Machine learning (ML) and artificial intelligence (AI) technologies are playing an increasingly critical role in the new era of precision psychiatry. Combining ML/AI with neuromodulation technologies can potentially provide explainable solutions in clinical practice and effective therapeutic treatment. Advanced wearable and mobile technologies also call for the new role of ML/AI for digital phenotyping in mobile mental health. In this review, we provide a comprehensive review of ML methodologies and applications by combining neuroimaging, neuromodulation, and advanced mobile technologies in psychiatry practice. We further review the role of ML in molecular phenotyping and cross-species biomarker identification in precision psychiatry. We also discuss explainable AI (XAI) and neuromodulation in a closed human-in-the-loop manner and highlight the ML potential in multi-media information extraction and multi-modal data fusion. Finally, we discuss conceptual and practical challenges in precision psychiatry and highlight ML opportunities in future research.

Toward biophysical markers of depression vulnerability

A major difficulty with treating psychiatric disorders is their heterogeneity: different neural causes can lead to the same phenotype. To address this, we propose describing the underlying pathophysiology in terms of interpretable, biophysical parameters of a neural model derived from the electroencephalogram. We analyzed data from a small patient cohort of patients with depression and controls. Using DCM, we constructed biophysical models that describe neural dynamics in a cortical network activated during a task that is used to assess depression state. We show that biophysical model parameters are biomarkers, that is, variables that allow subtyping of depression at a biological level. They yield a low dimensional, interpretable feature space that allowed description of differences between individual patients with depressive symptoms. They could capture internal heterogeneity/variance of depression state and achieve significantly better classification than commonly used EEG features. Our work is a proof of concept that a combination of biophysical models and machine learning may outperform earlier approaches based on classical statistics and raw brain data.

Strategies and foundations for scientific discovery in longitudinal studies of bipolar disorder

Bipolar disorder (BD) is a complex and dynamic condition with a typical onset in late adolescence or early adulthood followed by an episodic course with intervening periods of subthreshold symptoms or euthymia. It is complicated by the accumulation of comorbid medical and psychiatric disorders. The etiology of BD remains unknown and no reliable biological markers have yet been identified. This is likely due to lack of comprehensive ontological framework and, most importantly, the fact that most studies have been based on small nonrepresentative clinical samples with cross-sectional designs. We propose to establish large, global longitudinal cohorts of BD studied consistently in a multidimensional and multidisciplinary manner to determine etiology and help improve treatment. Herein we propose collection of a broad range of data that reflect the heterogenic phenotypic manifestations of BD that include dimensional and categorical measures of mood, neurocognitive, personality, behavior, sleep and circadian, life-story, and outcomes domains. In combination with genetic and biological information such an approach promotes the integrating and harmonizing of data within and across current ontology systems while supporting a paradigm shift that will facilitate discovery and become the basis for novel hypotheses.

Machine Learning Algorithm-Based Prediction Model for the Augmented Use of Clozapine with Electroconvulsive Therapy in Patients with Schizophrenia

The augmentation of clozapine with electroconvulsive therapy (ECT) has been an optimal treatment option for patients with treatment- or clozapine-resistant schizophrenia. Using data from the Research on Asian Psychotropic Prescription Patterns for Antipsychotics survey, which was the largest international psychiatry research collaboration in Asia, our study aimed to develop a machine learning algorithm-based substantial prediction model for the augmented use of clozapine with ECT in patients with schizophrenia in terms of precision medicine. A random forest model and least absolute shrinkage and selection operator (LASSO) model were used to develop a substantial prediction model for the augmented use of clozapine with ECT. Among the 3744 Asian patients with schizophrenia, those treated with a combination of clozapine and ECT were characterized by significantly greater proportions of females and inpatients, a longer duration of illness, and a greater prevalence of negative symptoms and social or occupational dysfunction than those not treated. In the random forest model, the area under the curve (AUC), which was the most preferred indicator of the prediction model, was 0.774. The overall accuracy was 0.817 (95% confidence interval, 0.793−0.839). Inpatient status was the most important variable in the substantial prediction model, followed by BMI, age, social or occupational dysfunction, persistent symptoms, illness duration > 20 years, and others. Furthermore, the AUC and overall accuracy of the LASSO model were 0.831 and 0.644 (95% CI, 0.615−0.672), respectively. Despite the subtle differences in both AUC and overall accuracy of the random forest model and LASSO model, the important variables were commonly shared by the two models. Using the machine learning algorithm, our findings allow the development of a substantial prediction model for the augmented use of clozapine with ECT in Asian patients with schizophrenia. This substantial prediction model can support further studies to develop a substantial prediction model for the augmented use of clozapine with ECT in patients with schizophrenia in a strict epidemiological context.

Improving the Applicability of AI for Psychiatric Applications through Human-in-the-loop Methodologies

OBJECTIVES

Machine learning (ML) and natural language processing have great potential to improve efficiency and accuracy in diagnosis, treatment recommendations, predictive interventions, and scarce resource allocation within psychiatry. Researchers often conceptualize such an approach as operating in isolation without much need for human involvement, yet it remains crucial to harness human-in-the-loop practices when developing and implementing such techniques as their absence may be catastrophic. We advocate for building ML-based technologies that collaborate with experts within psychiatry in all stages of implementation and use to increase model performance while simultaneously increasing the practicality, robustness, and reliability of the process.

METHODS

We showcase pitfalls of the traditional ML framework and explain how it can be improved with human-in-the-loop techniques. Specifically, we applied active learning strategies to the automatic scoring of a story recall task and compared the results to a traditional approach.

RESULTS

Human-in-the-loop methodologies supplied a greater understanding of where the model was least confident or had knowledge gaps during training. As compared to the traditional framework, less than half of the training data were needed to reach a given accuracy.

CONCLUSIONS

Human-in-the-loop ML is an approach to data collection and model creation that harnesses active learning to select the most critical data needed to increase a model's accuracy and generalizability more efficiently than classic random sampling would otherwise allow. Such techniques may additionally operate as safeguards from spurious predictions and can aid in decreasing disparities that artificial intelligence systems otherwise propagate.

Identifying neuroimaging biomarkers of major depressive disorder from cortical hemodynamic responses using machine learning approaches

Background

Early diagnosis of major depressive disorder (MDD) could enable timely interventions and effective management which subsequently improve clinical outcomes. However, quantitative and objective assessment tools for the suspected cases who present with depressive symptoms have not been fully established.

Methods

Based on a large-scale dataset (n = 363 subjects) collected with functional near-infrared spectroscopy (fNIRS) measurements during the verbal fluency task (VFT), this study proposed a data representation method for extracting spatiotemporal characteristics of NIRS signals, which emerged as candidate predictors in a two-phase machine learning framework to detect distinctive biomarkers for MDD. Supervised classifiers (e.g., support vector machine (SVM), k-nearest neighbors (KNN)) cooperated with cross-validation were implemented to evaluate the predictive capability of selected features in a training set. Another test set that was not involved in developing the algorithms enabled the independent assessment of the model's generalization.

Findings

For the classification with the optimal fusion features, the SVM classifier achieved the highest accuracy of 75.6% ± 4.7% in the nested cross-validation, and the correct prediction rate of 78.0% with a sensitivity of 75.0% and a specificity of 81.4% in the test set. Moreover, the multiway ANOVA test on clinical and demographic factors confirmed that twenty out of 39 optimal features were significantly correlated with the MDD-distinctive consequence.

Interpretation

The abnormal prefrontal activity of MDD may be quantified as diminished relative intensity and inappropriate activation timing of hemodynamic response, resulting in an objectively measurable biomarker for assessing cognitive deficits and screening MDD at the early stage.

Funding

This study was funded by NUS iHeathtech Other Operating Expenses (R-722-000-004-731).

Patient-specific connectomic models correlate with, but do not reliably predict, outcomes in deep brain stimulation for obsessive-compulsive disorder

Deep brain stimulation (DBS) of the ventral internal capsule/ventral striatum (VCVS) is an emerging treatment for obsessive-compulsive disorder (OCD). Recently, multiple studies using normative connectomes have correlated DBS outcomes to stimulation of specific white matter tracts. Those studies did not test whether these correlations are clinically predictive, and did not apply cross-validation approaches that are necessary for biomarker development. Further, they did not account for the possibility of systematic differences between DBS patients and the non-diagnosed controls used in normative connectomes. To address these gaps, we performed patient-specific diffusion imaging in 8 patients who underwent VCVS DBS for OCD. We delineated tracts connecting thalamus and subthalamic nucleus (STN) to prefrontal cortex via VCVS. We then calculated which tracts were likely activated by individual patients' DBS settings. We fit multiple statistical models to predict both OCD and depression outcomes from tract activation. We further attempted to predict hypomania, a VCVS DBS complication. We assessed all models' performance on held-out test sets. With this best-practices approach, no model predicted OCD response, depression response, or hypomania above chance. Coefficient inspection partly supported prior reports, in that capture of tracts projecting to cingulate cortex was associated with both YBOCS and MADRS response. In contrast to prior reports, however, tracts connected to STN were not reliably correlated with response. Thus, patient-specific imaging and a guideline-adherent analysis were unable to identify a tractographic target with sufficient effect size to drive clinical decision-making or predict individual outcomes. These findings suggest caution in interpreting the results of normative connectome studies.

Natural language processing in clinical neuroscience and psychiatry: A review

Natural language processing (NLP) is rapidly becoming an important topic in the medical community. The ability to automatically analyze any type of medical document could be the key factor to fully exploit the data it contains. Cutting-edge artificial intelligence (AI) architectures, particularly machine learning and deep learning, have begun to be applied to this topic and have yielded promising results. We conducted a literature search for 1,024 papers that used NLP technology in neuroscience and psychiatry from 2010 to early 2022. After a selection process, 115 papers were evaluated. Each publication was classified into one of three categories: information extraction, classification, and data inference. Automated understanding of clinical reports in electronic health records has the potential to improve healthcare delivery. Overall, the performance of NLP applications is high, with an average F1-score and AUC above 85%. We also derived a composite measure in the form of Z-scores to better compare the performance of NLP models and their different classes as a whole. No statistical differences were found in the unbiased comparison. Strong asymmetry between English and non-English models, difficulty in obtaining high-quality annotated data, and train biases causing low generalizability are the main limitations. This review suggests that NLP could be an effective tool to help clinicians gain insights from medical reports, clinical research forms, and more, making NLP an effective tool to improve the quality of healthcare services.

Physiologically informed neuromodulation

The rapid evolution of neuromodulation techniques includes an increasing amount of research into stimulation paradigms that are guided by patients' neurophysiology, to increase efficacy and responder rates. Treatment personalisation and target engagement have shown to be effective in fields such as Parkinson's disease, and closed-loop paradigms have been successfully implemented in cardiac defibrillators. Promising avenues are being explored for physiologically informed neuromodulation in psychiatry. Matching the stimulation frequency to individual brain rhythms has shown some promise in transcranial magnetic stimulation (TMS). Matching the phase of those rhythms may further enhance neuroplasticity, for instance when combining TMS with electroencephalographic (EEG) recordings. Resting-state EEG and event-related potentials may be useful to demonstrate connectivity between stimulation sites and connected areas. These techniques are available today to the psychiatrist to diagnose underlying sleep disorders, epilepsy, or lesions as contributing factors to the cause of depression. These technologies may also be useful in assessing the patient's brain network status prior to deciding on treatment options. Ongoing research using invasive recordings may allow for future identification of mood biomarkers and network structure. A core limitation is that biomarker research may currently be limited by the internal heterogeneity of psychiatric disorders according to the current DSM-based classifications. New approaches are being developed and may soon be validated. Finally, care must be taken when incorporating closed-loop capabilities into neuromodulation systems, by ensuring the safe operation of the system and understanding the physiological dynamics. Neurophysiological tools are rapidly evolving and will likely define the next generation of neuromodulation therapies.

Prediction and Prevention in the Clinical High-Risk for Psychosis Paradigm: A Review of the Current Status and Recommendations for Future Directions of Inquiry

Prediction and prevention of negative clinical and functional outcomes represent the two primary objectives of research conducted within the clinical high-risk for psychosis (CHR-P) paradigm. Several multivariable "risk calculator" models have been developed to predict the likelihood of developing psychosis, although these models have not been translated to clinical use. Overall, less progress has been made in developing effective interventions. In this paper, we review the existing literature on both prediction and prevention in the CHR-P paradigm and, primarily, outline ways in which expanding and combining these paths of inquiry could lead to a greater improvement in individual outcomes for those most at risk.

Machine Learning Prediction of Treatment Outcome in Late-Life Depression

Background: Recent evidence suggests that integration of multi-modal data improves performance in machine learning prediction of depression treatment outcomes. Here, we compared the predictive performance of three machine learning classifiers using differing combinations of sociodemographic characteristics, baseline clinical self-reports, cognitive tests, and structural magnetic resonance imaging (MRI) features to predict treatment outcomes in late-life depression (LLD). Methods: Data were combined from two clinical trials conducted with depressed adults aged 60 and older, including response to escitalopram (N = 32, NCT01902004) and Tai Chi (N = 35, NCT02460666). Remission was defined as a score of 6 or less on the 24-item Hamilton Rating Scale for Depression (HAMD) at the end of 24 weeks of treatment. Features subsets were constructed from baseline sociodemographic and clinical features, gray matter volumes (GMVs), or both. Three classification algorithms were compared: (1) Support Vector Machine-Radial Bias Function (SVMRBF), (2) Random Forest (RF), and (3) Logistic Regression (LR). A repeated 5-fold cross-validation approach with a wrapper-based feature selection method was used for model fitting. Model performance metrics included Area under the ROC Curve (AUC) and Matthews correlation coefficient (MCC). Cross-validated performance significance was tested by permutation analysis. Classifiers were compared by Cochran's Q and post-hoc pairwise comparisons using McNemar's Chi-Square test with Bonferroni correction. Results: For the RF and SVMRBF algorithms, the combined feature set outperformed the clinical and GMV feature sets with a final cross-validated AUC of 0.83 ± 0.11 and 0.80 ± 0.11, respectively. Both classifiers passed permutation analysis. The LR algorithm performed best using GMV features alone (AUC 0.79 ± 0.14) but failed to pass permutation analysis using any feature set. Performance of the three classifiers differed significantly for all three features sets. Important predictive features of treatment response included anterior and posterior cingulate volumes, depression characteristics, and self-reported health-related quality scores. Conclusion: This preliminary exploration into the use of ML and multi-modal data to identify predictors of general treatment response in LLD indicates that integration of clinical and structural MRI features significantly increases predictive capability. Identified features are among those previously implicated in geriatric depression, encouraging future work in this arena.