Predicting new onset thought disorder in early adolescence with optimized deep learning implicates environmental-putamen interactions

Abstract

Background: Thought disorder (TD) is a sensitive and specific marker of risk for schizophrenia onset. Specifying factors that predict TD onset in adolescence is important to early identification of youth at risk. However, there is a paucity of studies prospectively predicting TD onset in unstratified youth populations.

Study design: We used deep learning optimized with artificial intelligence (AI) to analyze 5,777 multimodal features obtained at 9-10 years from youth and their parents in the ABCD study, including 5,014 neural metrics, to prospectively predict new onset TD cases at 11-12 years. The design was replicated for all prevailing TD cases at 11-12 years.

Study results: Optimizing performance with AI, we were able to achieve 92% accuracy and F1 and 0.96 AUROC in prospectively predicting the onset of TD in early adolescence. Structural differences in the left putamen, sleep disturbances and the level of parental externalizing behaviors were specific predictors of new onset TD at 11-12 yrs, interacting with low youth prosociality, the total parental behavioral problems and parent-child conflict and whether the youth had already come to clinical attention. More important predictors showed greater inter-individual variability.

Conclusions: This study provides robust person-level, multivariable signatures of early adolescent TD which suggest that structural differences in the left putamen in late childhood are a candidate biomarker that interacts with psychosocial stressors to increase risk for TD onset. Our work also suggests that interventions to promote improved sleep and lessen parent-child psychosocial stressors are worthy of further exploration to modulate risk for TD onset.

Keywords: Adolescence; Deep Learning; Environmental; New Onset; Putamen; Thought Disorder.

Figure 1:. Construction of participant case samples

Steps used to form the study sample are displayed. After inclusion criteria were applied, the sample was randomly partitioned into training and test sets. Subsequently, samples for each experiment were formed as described in Preparation of predictive targets and Construction of participant case samples.

Figure 2:. Receiver Operating Characteristic curves for deep learning optimized with Integrated Evolutionary Learning in predicting cases of thought disorder in early adolescence

Receiver Operating Characteristic (ROC) curves are displayed for the most accurate model predicting TD cases obtained with deep learning optimized with Integrated Evolutionary Learning for a) new onset cases; and b) all prevailing cases at 11–12 years of age. These ROC curves correspond to performance statistics shown in Table 5.

Figure 3:. Person-level predictor importances of new onset and prevailing cases of Thought Disorder

Summary plots are presented of the importance of each final predictor (computed with the Shapley Additive Explanations technique) on an individual subject level to predicting a) TD with new onset at 11–12 yrs and b) all prevailing cases of TD at 11–12 yrs with features collected at 9–10 yrs. The color gradient represents the original value of each feature where red = high and blue = low. Discrete (binary) features appear as red or blue, while continuous features appear as a color gradient.