Machine learning for suicidology: A practical review of exploratory and hypothesis-driven approaches

Cox, Christopher R.; Moscardini, Emma H.; Cohen, Alex S.; Tucker, Raymond P.

doi:10.1016/j.cpr.2020.101940

Cited by 27 publications

(25 citation statements)

References 60 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…AI is used in a wide range of applications within mental health, notably within clinical research settings where data are used to aid in understanding the nature of diagnoses and to improve diagnostic accuracy (for reviews see Shatte et al, 2019;Su et al, 2020;Thieme et al, 2020), as well as in making complex and potentially lifesaving de-cisions (e.g., in suicidology -for review see Cox et al, 2020). Acoustic measurements of speech have been analyzed in automated applications for detecting Mild Cognitive Impairment and dementia (Roark et al, 2011;König et al, 2015), as well as serious mental illness and depression (McGinnis et al, 2019).…”

Section: Introductionmentioning

confidence: 99%

Safeguarding against spurious AI-based predictions: The case of automated verbal memory assessment

Chandler¹,

Foltz²,

Cohen³

et al. 2021

Proceedings of the Seventh Workshop on Computational Linguistics and Clinical Psychology: Improving Access

View full text Add to dashboard Cite

A growing amount of psychiatric research incorporates machine learning and natural language processing methods, however findings have yet to be translated into actual clinical decision support systems. Many of these studies are based on relatively small datasets in homogeneous populations, which has the associated risk that the models may not perform adequately on new data in real clinical practice. The nature of serious mental illness is that it is hard to define, hard to capture, and requires frequent monitoring, which leads to imperfect data where attribute and class noise are common. With the goal of an effective AI-mediated clinical decision support system, there must be computational safeguards placed on the models used in order to avoid spurious predictions and thus allow humans to review data in the settings where models are unstable or bound not to generalize. This paper describes two approaches to implementing safeguards: (1) the determination of cases in which models are unstable by means of attribute and class based outlier detection and (2) finding the extent to which models show inductive bias. These safeguards are illustrated in the automated scoring of a story recall task via natural language processing methods. With the integration of human-in-the-loop machine learning in the clinical implementation process, incorporating safeguards such as these into the models will offer patients increased protection from spurious predictions.

show abstract

Section: Introductionmentioning

confidence: 99%

Safeguarding against spurious AI-based predictions: The case of automated verbal memory assessment

Chandler¹,

Foltz²,

Cohen³

et al. 2021

Proceedings of the Seventh Workshop on Computational Linguistics and Clinical Psychology: Improving Access

View full text Add to dashboard Cite

show abstract

Section: Introductionmentioning

confidence: 99%

Proceedings of the Seventh Workshop on Computational Linguistics and Clinical Psychology: Improving Access

2021

View full text Add to dashboard Cite

Recently, research on mental health conditions using public online data, including Reddit, has surged in NLP and health research but has not reported user characteristics, which are important to judge generalisability of findings. This paper shows how existing NLP methods can yield information on clinical, demographic, and identity characteristics of almost 20K Reddit users who self-report a bipolar disorder diagnosis. This population consists of slightly more feminine-than masculinegendered mainly young or middle-aged USbased adults who often report additional mental health diagnoses, which is compared with general Reddit statistics and epidemiological studies. Additionally, this paper carefully evaluates all methods and discusses ethical issues. ReferencesWasim Ahmed, Peter A. Bath, and Gianluca Demartini.2017. Using Twitter as a data source: an overview of ethical, legal and methodological challenges. In Kandy Woodfield, editor, The Ethics of Online Research, pages 79-107. Emerald Books.

show abstract

“…Suicide attempts are rare, but the consequences are dire. Machine learning (ML) has therefore gained traction because of the potential to have high impact in increasing sensitivity to identify imminent suicidal behaviors (Cox, Moscardini, Cohen, & Tucker, 2020). STBs appear across mental health disorders, which is one reason why STBs are so difficult to predict.…”

Section: Introductionmentioning

confidence: 99%

“…One advantage of ML is the ability to search a large set of models and then pick the one that increases model performance, which can thus yield more precision in characterizing the variability in psychopathology (Bzdok & Meyer-Lindenberg, 2018). There are several challenges that come with a method that aims to increase predictive accuracy on datasets with a large number of variables without a strong theoretical basis (Cox et al, 2020). One challenge is interpretation of the output – the independent variables that are considered important predictors of STBs.…”

Section: Introductionmentioning

confidence: 99%

“…One challenge is interpretation of the output – the independent variables that are considered important predictors of STBs. Specifically, it can be challenging to interpret the clinical significance of each variable that is considered “important” within a particular dataset because more often than not they are not based on pre-specified hypotheses or theory (Cox et al, 2020). Given the nascent stage of biological predictors of STBs, we aimed to use both hypothesis-driven and data-driven approaches to characterize organizational properties of the whole brain in concert with psychological and environmental factors that previous research have shown to be important to identify the combination of factors that predict future severity of SI.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Brain Connectomics and Severity of Internalizing Symptoms in Early Adolescence Predict Severity of Suicidal Ideation in Later Adolescence

Kirshenbaum

Chahal

Ho³

et al. 2020

Preprint

View full text Add to dashboard Cite

Suicidal ideation (SI) typically emerges during adolescence but is challenging to predict. Given the consequences of SI, it is important to identify neurobiological and psychological predictors of SI in adolescents in order to improve suicide prevention strategies. We used graph theoretical methods to examine whether, and which, local properties of functional brain topology in early adolescence can predict severity of subsequent SI. In 109 participants (61 female) ages 9-13 years, we assessed clinical and participant characteristics and obtained resting-state fMRI data (baseline). We examined inter-regional functional connectedness across 250 brain regions by computing measures of nodal interconnectedness: local efficiency, eigenvector centrality, nodal degree, within-module z-score, and participation coefficient. A LASSO regression identified a linear combination of the most important predictors of SI severity measured approximately 4 years later. The LASSO analysis identified a combination of 8 predictors of future SI severity (R2=.23) including internalizing symptom severity as the largest predictor (β=.13). The remaining predictors were brain-based including nodal degree of the inferior frontal gyrus (β=-.08), precentral gyrus (β=-.07), fusiform gyrus (β=.05), and inferior temporal gyrus (β=.03), within-module degree of the substantia nigra (β=.06) and inferior parietal lobe (β=-.03), and eigenvector centrality of the subgenual cingulate gyrus (β=.03). Our findings suggest that along with early symptoms of internalizing behaviors, network properties of the brain in early adolescence, a period when SI may not be clinically evident, are markers of vulnerability for SI severity later in adolescence. Research is needed to validate the clinical utility of these markers as predictors of suicidal thoughts.

show abstract

Machine learning for suicidology: A practical review of exploratory and hypothesis-driven approaches

Cited by 27 publications

References 60 publications

Safeguarding against spurious AI-based predictions: The case of automated verbal memory assessment

Safeguarding against spurious AI-based predictions: The case of automated verbal memory assessment

Proceedings of the Seventh Workshop on Computational Linguistics and Clinical Psychology: Improving Access

Brain Connectomics and Severity of Internalizing Symptoms in Early Adolescence Predict Severity of Suicidal Ideation in Later Adolescence

Contact Info

Product

Resources

About