Towards a Network Control Theory of Electroconvulsive Therapy Response

Abstract

Electroconvulsive Therapy (ECT) is arguably the most effective intervention for treatment-resistant depression. While large interindividual variability exists, a theory capable of predicting individual response to ECT remains elusive. To address this, we posit a quantitative, mechanistic framework of ECT response based on Network Control Theory (NCT). Then, we empirically test our approach and employ it to predict ECT treatment response. To this end, we derive a formal association between Postictal Suppression Index (PSI) -an ECT seizure quality index -and whole-brain modal and average controllability, NCT metrics based on white matter brain network architecture, respectively. Exploiting the known association of ECT response and PSI, we then hypothesized an association between our controllability metrics and ECT response mediated by PSI. We formally tested this conjecture in N=50 depressive patients undergoing ECT. We show that whole-brain controllability metrics based on pre-ECT structural connectome data predict ECT response in accordance with our hypotheses. In addition, we show the expected mediation effects via PSI. Importantly, our theoretically motivated metrics are at least on par with extensive machine learning models based on pre-ECT connectome data. In summary, we derived and tested a control-theoretic framework capable of predicting ECT response based on individual brain network architecture. It makes testable, quantitative predictions regarding individual therapeutic response, which are corroborated by strong empirical evidence. Our work might constitute a starting point for a comprehensive, quantitative theory of personalized ECT interventions rooted in control theory.

Network controllability measures of subnetworks: implications for neurosciences

“…Thus, after normalization, both A and A s have one eigenvalue equal to 1 and one bulk of small eigenvalues, where the bulk of A s is an order of 1/µ 1 larger than that of A. This fact nicely explains the mainly negative (positive) trends we observed in ∆MC (∆Ac) since, as we showed in our recent work [18], the average (over all network nodes) AC and MC have the simple expressions as shown in equations ( 6) and (7).…”

“…In neuroscience, network control theory is increasingly used [10,18,19,21,39], for good reasons, to study neural, biological, and psychological constructs. It relates fundamental theory-driven results from controls literature to the study of networks that are the natural points of interest in neurosciences [14].…”

“…For this and similar questions, assuming linear temporal dynamics, network control theory provides a set of measures that mathematically quantify the relevance of the nodes [17]. Specifically, average-, and modal-controllability are among the most frequently-used measures in neurosciences to study neural [9,10,[18][19][20] as well as psychological symptom dynamics [21,22] and are suggested to measure the average ability of nodes to affect network dynamics (see Karrer et al for a detailed mathematical introduction [23]).…”

Formalizing psychological interventions through network control theory

“…where 𝑇 and 𝜌 are free parameters quantifying the time to reach from 𝑥 0 to 𝑥 𝑇 and the relative importance of cost terms in equation ( 5). Following (Gu et al, 2015;Hahn, Jamalabadi, Nozari, et al, 2021), we define the step size to 0.001, 𝑇 = 1, and 𝜌 = 1. To solve the optimal control equations ( 5) and ( 6), we use a customized version of the code that is used elsewhere to study the brain as well as psychological dynamics (Gu et al, 2017;Hahn, Jamalabadi, Emden, et al, 2021;Jamalabadi et al, 2022).…”