Decoding Depression Severity From Intracranial Neural Activity

Predictive neuromodulation of cingulo-frontal neural dynamics in major depressive disorder using a brain-computer interface system: A simulation study

“…Specifically, without DBS treatment (I dbs = 0), the simulated MDD state (f D = 1.25) has lower vACC θ power and higher vACC β + γ power than the simulated healthy state (f D = 1) (Figure 1B, left panel). This is in line with recent findings in human intracranial electroencephalography (iEEG) studies that in some patients, cingulate θ power can decrease (Sani et al, 2018;Bijanzadeh et al, 2022) and β/γ power can increase (Sani et al, 2018;Bijanzadeh et al, 2022;Xiao et al, 2023) when mood symptoms become worse. This is consistent with the DBS treatment effects where cingulate θ power can increase (Smith et al, 2022) and β/γ power can decrease after DBS (i.e., improved mood symptoms) (Smart et al, 2018).…”

“…The original vACC-dlPFC neural mass model focuses on investigating the temporal dynamics of raw neural activity , , , and demonstrates that the model shows interesting non-linear dynamics that are consistent with clinical findings (Ramirez-Mahaluf et al, 2017 ). However, recent clinical findings on neural signatures of mood symptoms have shown that spectral power dynamics—especially in the θ (3–7 Hz) and β+low γ (13–50 Hz) bands—are key features that are related to mood symptoms (Rao et al, 2018 ; Sani et al, 2018 ; Smart et al, 2018 ; Scangos et al, 2021a ; Bijanzadeh et al, 2022 ; Smith et al, 2022 ; Xiao et al, 2023 ). Therefore, we make two adjustments to the original vACC-dlPFC model such that we can simulate spectral power dynamics that are in line with the above clinical findings.…”

“…Each averaged spectral response changes non-linearly and di erently as the DBS amplitude varies. Xiao et al, 2023). Therefore, we make two adjustments to the original vACC-dlPFC model such that we can simulate spectral power dynamics that are in line with the above clinical findings.…”

“…Therefore, we adopt an established computational vACC-dlPFC model (Ramirez-Mahaluf et al, 2017) to simulate the neural activity in MDD. We especially adjust the original model such that the output spectral power dynamics are in line with the most recent findings in spectral signatures of mood symptoms (Rao et al, 2018;Sani et al, 2018;Smart et al, 2018;Scangos et al, 2021a;Smith et al, 2022;Xiao et al, 2023). We then use the adjusted model as a simulation testbed to evaluate offline dynamic system identification methods (see Section 2.2.2) and online neuromodulation techniques (see Sections 2.2.3 and 2.2.4).…”

“…First, MDD is a complex neuropsychiatric disorder that likely involves the dysfunction of a distributed network consisting of multiple limbic and frontal regions (Mayberg, 1997 ; Drevets, 2001 ; Ramirez-Mahaluf et al, 2017 ; Williams, 2017 ). Moreover, within the limbic-frontal network, altered neural spectral activity at different frequency bands—especially the θ band (3–7 Hz) and β+low γ band (30–50 Hz)—has been observed in different mood symptom states (Sani et al, 2018 ; Bijanzadeh et al, 2022 ; Xiao et al, 2023 ) and after DBS (Rao et al, 2018 ; Smart et al, 2018 ; Smith et al, 2022 ). This means that using a scalar neural biomarker computed from a single brain site and a single frequency band (Scangos et al, 2021a ) may not be sufficient for achieving precise MDD symptom regulation.…”