Search for: All records

This paper proposes a method to learn ap- proximations of missing Ordinary Differential Equations (ODEs) and states in physiological models where knowl- edge of the system’s relevant states and dynamics is in- complete. The proposed method augments known ODEs with neural networks (NN), then trains the hybrid ODE-NN model on a subset of available physiological measurements (i.e., states) to learn the NN parameters that approximate the unknown ODEs. Thus, this method can model an ap- proximation of the original partially specified system sub- ject to the constraints of known biophysics. This method also addresses the challenge of jointly estimating physio- logical states, NN parameters, and unknown initial condi- tions during training using recursive Bayesian estimation. We validate this method using two simulated physiolog- ical systems, where subsets of the ODEs are assumed to be unknown during the training and test processes. The proposed method almost perfectly tracks the ground truth in the case of a single missing ODE and state and performs well in other cases where more ODEs and states are missing. This performance is robust to input signal per- turbations and noisy measurements. A critical advantage of the proposed hybrid methodology over purely data-driven methods is the incorporation of the ODE structure in the model, which allows one to infer unobserved physiological states. The ability to flexibly approximate missing or inac- curate components in ODE models improves a significant modeling bottleneck without sacrificing interpretability. 

In this article, we propose a theoretical model leveraging the analogy between fluid and electric variables to investigate the relation among aqueous humor (AH) circulation and drainage and intraocular pressure (IOP), the principal established risk factor of severe neuropathologies of the optic nerve such as glaucoma. IOP is the steady-state result of the balance among AH secretion (AHs), circulation (AHc), and drainage (AHd). AHs are modeled as a given volumetric flow rate electrically corresponding to an input current source. AHc is modeled by the series of two linear hydraulic conductances (HCs) representing the posterior and anterior chambers. AHd is modeled by the parallel of three HCs: a linear HC for the conventional adaptive route (ConvAR), a nonlinear HC for the hydraulic component of the unconventional adaptive route (UncAR), and a nonlinear HC for the drug-dependent component of the UncAR. The proposed model is implemented in a computational virtual laboratory to study the value attained by the IOP under physiological and pathological conditions. Simulation results (i) confirm the conjecture that the UncAR acts as a relief valve under pathological conditions, (ii) indicate that the drug-dependent AR is the major opponent to IOP increase in the case of elevated trabecular meshwork resistance, and (iii) support the use of the model as a quantitative tool to complement in vivo studies and help design and optimize medications for ocular diseases. 

This study investigated the heterogeneity of ocular hemodynamic biomarkers in early open angle glaucoma (OAG) patients and healthy controls of African (AD) and European descent (ED). Sixty OAG patients (38 ED, 22 AD) and 65 healthy controls (47 ED, 18 AD) participated in a prospective, cross-sectional study assessing: intraocular pressure (IOP), blood pressure (BP), ocular perfusion pressure (OPP), visual field (VF) and vascular densities (VD) via optical coherence tomography angiography (OCTA). Comparisons between outcomes were adjusted for age, diabetes status and BP. VF, IOP, BP and OPP were not significantly different between OAG subgroups or controls. Multiple VD biomarkers were significantly lower in OAG patients of ED (p < 0.05) while central macular VD was lower in OAG patients of AD vs. OAG patients of ED (p = 0.024). Macular and parafoveal thickness were significantly lower in AD OAG patients compared to those of ED (p = 0.006–0.049). OAG patients of AD had a negative correlation between IOP and VF index (r = −0.86) while ED patients had a slightly positive relationship (r = 0.26); difference between groups (p < 0.001). Age-adjusted OCTA biomarkers exhibit significant variation in early OAG patients of AD and ED.