Vasilis Z. Marmarelis, Ph.D. - US grants

model design /development; computer simulation; biomedical resource; biomedical facility;

The Biomedical Simulations Resource (BMSR) at the University of Souther California was established in September 1985 through Grant No. RR-01861 from the National Center for Research Resources of the NIH. The BMSR is dedicated to the advancement of the state of the art in biomedical modeling and simulation through Technological Research and Development (Core) and Collaborative Research projects. It also seeks the dissemination of this knowledge and related software through Service, Training and Dissemination activities aimed at the biomedical community at large. The latter activities include development and distribution of modeling/simulation software, short-courses, workshops and publication of research volumes-in addition to publications in the open literature. The emphasis of the Core Research projects is on development of advanced modeling and simulation methodologies and their application to physiological systems for the advancement of knowledge and improvement of clinical practice. The challenging cases of non-linear, non-stationary, sparse-data, feedback and physiological control systems, as well as the modeling/simulation of complex biomedical systems, constitute the focal points of this effort. The methodologies developed by the Core projects are at the cutting edge of research in this area. Their effective study is made possible by the computing facilities and human resources that are brought together in the BMSR. Pivotal applications of these methodologies cover a variety of physiological domains, including pharmacokinetics, cardiorespiratory, neural and sensory systems. Additional applications, serving as testing grounds and dissemination means for the developed methodologies, are made in our Collaborative projects. Fully developed and tested methodologies are implemented in user-friendly software packages and made available to the biomedical community at large. The Service, Training and Dissemination activities of the BMSR seek to promote the exchange of knowledge and technology in the area of biomedical modeling and simulation. They also aim at creating an interactive community of investigators, which can provide the multi-disciplinary scientific/technological impetus needed to transcend the barriers imposed by the complexity of the subject. The ultimate objective of the BMSR is to facilitate transfer of simulation science and technology to the clinical environment for the improvement of our health care system.

One of this year's Advance Short Course, entitled "Nonlinear System Modeling and Data Analysis Using LYSIS Version 6.2", was conducted by Professor Marmarelis and Dr. Courellis and held on May 30-31, 1998 at the USC School of Engineering. This course provided the 20 attendees with intensive hands-on instruction, methodological background and the key practical considerations in analyzing real data. Class demonstrations of examples in utilizing simulated and real data were given, followed by extensive interactions with the participants. Very favorable comments were received by the participants about the impact of the Short Course and their work. Representatives from two research groups have subsequently sent us data in pursuit of a future collaboration.

technology /technique development; nervous system; statistics /biometry; informatics; eye; biomedical resource; bioengineering /biomedical engineering;

One of this year's Advance Short Course, entitled "Nonlinear System Modeling and Data Analysis Using LYSIS Version 6.2", was conducted by Professor Marmarelis and Dr. Courellis and held on May 30-31, 1998 at the USC School of Engineering. This course provided the 20 attendees with intensive hands-on instruction, methodological background and the key practical considerations in analyzing real data. Class demonstrations of examples in utilizing simulated and real data were given, followed by extensive interactions with the participants. Very favorable comments were received by the participants about the impact of the Short Course and their work. Representatives from two research groups have subsequently sent us data in pursuit of a future collaboration.

This project generated 14 publications (6 in refereed journals, one book chapter, and 7 in conference proceedings) during the past year, in addition to 7 papers that have been submitted for publication in refereed journals. We can define six main areas of research progress: (1) nonlinear modeling of physiological systems; (2) nonstationary modeling of nonlinear physiological systems; (3) nonlinear modeling with point-process input data; (4) decomposition of nonlinear feedback via adaptive estimation; (5) nonlinear parametric models from nonparametric measurements; and (6) laser-induced fluorescence spectroscopy Of particular importance are the general methodologies that yield compact nonlinear and nonstationary models within the practical limitations of experimental investigations, solving a long-standing problem of critical importance for physiological system modeling. Pilot applications to experimental data from the rat kidney, spider mechanoreceptors and the rabbit hippocampus have yielded new insights into the physiological mechanisms of these systems.

EIA-0130883-Theodore W. Berger-University of Southern California-Title: Neurobiological Nonlinear Dynamics for Biomimetic Signal Processing-Title-The fundamental goal of the proposed research is to derived a new generation of temporal and spatio-temporal pattern recognition systems based on the nonlinear dynamics, network architecture, and synaptic plasticity properties of the hippocampus, a cortical brain system responsible for the formation of new pattern recognition memories. From a neurobiological perspective, the proposed experimental/modeling work promises to generate (1) first-characterizations of high-order nonlinearities of cortical brain tissue, i.e., predictive models of the input/output transformations in spatio-temporal activity performed by individual hippocampal neurons, and to (2) investigate the increasingly likely possibility of dynamic neural "learning rules", i.e., requisite conditions for the induction of synaptic plasticity that depend on the past history of activity. In addition, the proposed research will investigate (3) the role of known hippocampal network topology in neurobiological signal processing and hierarchical feature extraction. From a theoretical/computational perspective, the proposed work is designed to (4) develop novel methodologies essential for characterizing nonlinearities of neurobiological systems, as well as to (5) further expand a newly developed paradigm for biologically realistic neural system modeling (the "dynamic synapse neural network architecture") that has already demonstrated a heretofore unmatched capability for identifying optimal feature sets for temporally and spatio-temporally coded information.

The development of our general methodology has been rooted in the rigorous mathematical framework of the Volterra-Wiener approach to nonlinear system analysis and its various extensions over the last 35 years [9-16]. This methodological framework was selected because: (a) it is mathematically rigorous and allows practical identification of the system dynamics from input-output data without requiring prior knowledge of the internal organization of the system; (b) it is applicable to a very broad class of nonlinear dynamic biomedical systems (i.e. those with finite memory); (c) it yields models that represent the system function under broad (natural) operating conditions and allows predictions for arbitrary input signals (within the experimental range of frequencies/amplitudes); (d) it is robust in the presence of noise commonly found in biomedical time-series data. This remarkable set of attributes has been validated over the last 35 years. The development of practical methods that achieve high estimation accuracy with modest experimental and computational requirements has been achieved for single-input/single-output and multi-input/multi-output (MIMO) systems [14-60], although certain challenges remain with regard to compactness and interpretability of large-scale MIMO models, as well as the modeling of multi-variable nested-loop configurations. These issues represent the next generation of modeling challenges (as formulated in the Specific Aims) and attain rising importance as we begin to recognize the potential utility of these models for clinical purposes.

Model-based cerebrovascular markers extracted from hemodynamic data for non-invasive, portable and inexpensive diagnosis of MCI or mild AD and prediction of disease progression PROJECT SUMMARY The goal of the proposed multi-PI project is to establish proof of concept for the utility of a new class of cerebrovascular markers that may aid in the improved diagnosis and prediction of disease progression in Mild Cognitive Impairment (MCI) and mild Alzheimer's disease (AD). The means for obtaining these markers are non-invasive, inexpensive and portable, so that they can be used for screening in a primary-care setting. The scientific rationale for this new class of cerebrovascular markers is provided by the recent promising results of our group and the mounting evidence of a strong correlation between MCI/AD and cerebrovascular dysregulation. A recently published retrospective study on a large cohort of 1,171 subjects from the ADNI database utilized multi-factorial data-driven analysis to assess the relation between MCI/AD disease progression and commonly used biomarkers (obtained from MRI/PET and plasma/CSF) and concluded that cerebrovascular dysregulation is the earliest and strongest pathologic factor associated with AD progression, corroborating the hypothesis of cerebrovascular dysregulation. Quantification of cerebrovascular dysregulation in that large-cohort study was achieved through analysis of ASL-MRI data of cerebral perfusion. We propose instead to explore a novel integrative dynamic modeling approach that analyzes the cerebral hemodynamics of persons with no cognitive impairment and MCI/AD patients with a methodology that yields input- output predictive models of the dynamic relationships between changes in beat-to-beat cerebral blood flow velocity (via Transcranial Doppler) or cerebral tissue oxygenation (via Near Infrared Spectroscopy) in response to changes in arterial blood pressure and end-tidal CO2 data. The obtained data-based models are subsequently used to compute markers of the dynamics of cerebrovascular regulation. Initial results of the advocated approach have achieved statistically significant delineation between 46 MCI patients and 20 age-matched controls on the basis of a model-based marker of dynamic vasomotor reactivity (DVR). Evaluation of the DVR marker against established MRI-based and PET-based biomarkers, as well as neuropsychological test data, from the larger cohort of the proposed project offers the promise of portable, non-invasive, inexpensive and sensitive means for detecting cerebrovascular dysregulation at the early stages of MCI or mild AD, and monitoring disease progression. Important co-variates of this study include age, gender, education, ApoE genotype, site and amyloid burden.

Urgent Competitive Revision Supplement to the multi-PI award RO1AG058162 entitled: Model-based cerebrovascular markers extracted from hemodynamic data for non-invasive, portable and inexpensive diagnosis of MCI or mild AD and prediction of disease progression PROJECT SUMMARY The goal of the proposed Urgent Competitive Revision Supplement to the current multi-PI award RO1AG058162 is to expand the scope of the current program to include aspects of cardio-respiratory regulation of cerebral perfusion in a subset of volunteers from our current cohort (30 AD patients, 30 MCI patients and 30 cognitively-normal controls) as well as in 30 newly recruited Covid-recovered patients in order to investigate the cardio-respiratory regulation in MCI and AD patients, as well as the effect of Covid-19 on the regulation of cardio-respiratory control and cerebral perfusion. The latter issue has attained urgent clinical importance during the ongoing Covid-19 pandemic because of the observed dysfunction of the fundamental cardio-respiratory chemoreflex that appeared unable to restore the homeostatic balance in some severe cases of Covid-19 presenting very low blood oxygen saturation without the normally expected tachypnea (termed tentatively ?silent hypoxemia?). This proposed expansion of the scope of the current multi-PI program will further enhance the main objective regarding the potential utility of a new class of cerebrovascular markers for the improved diagnosis and prediction of disease progression in Mild Cognitive Impairment (MCI) and mild Alzheimer's Disease (AD). The means for obtaining these markers are non-invasive, inexpensive and portable, so that they can be used for screening in a primary-care setting. The scientific rationale for this new class of cerebrovascular markers is provided by recent promising results of our group and the mounting evidence of a strong correlation between MCI/AD and cerebrovascular dysregulation in the work of many others, which suggest that cerebrovascular dysregulation is the earliest and strongest pathologic factor associated with AD progression, corroborating the hypothesis of cerebrovascular dysregulation. The current research program and the proposed expansion of its scope will achieve reliable quantification of cerebrovascular dysregulation through our novel integrative approach of predictive dynamic modeling that analyzes the cerebral hemodynamics and cardio-respiratory regulation through the use of input-output predictive models of the dynamic relationships between changes in beat-to-beat cerebral blood flow velocity or cerebral tissue oxygenation in response to changes in arterial blood pressure, end-tidal CO2 data, blood oxygen saturation, heart rate and (with the expanded scope of the proposed Revision Supplement) changes in respiratory rate, ventilation and inhaled gases (O2 and CO2). The obtained data-based models are subsequently used to compute markers of the dynamics of cerebrovascular regulation. These model-based markers will be evaluated against established MRI-based and PET-based biomarkers, as well as neuropsychological test data, offering the promise of non-invasive, inexpensive and sensitive means for detecting cerebrovascular dysregulation and (with the proposed Supplement) cardio-respiratory dysregulation at the early stages of MCI or mild AD, as well as in many severe Covid-19 cases with the puzzling clinical presentation of ?silent hypoxemia? that is recognized as high mortality risk for hospitalized severe Covid-19 cases.