New Acceleration Fund Projects 2023-25

We propose a new collaboration between the labs of PI Joshua Jacobs (Biomedical Engineering) and PI Christopher Baldassano (Psychology) to study the neural mechanisms underlying naturalistic memory. Neuroscientists that use conventional lab-based experiments have struggled to understand why and how everyday memory can succeed or fail, largely due to the coarse traditional tools used to measure brain activity and memory performance. Traditional memory paradigms that use lists of random words or pictures fail to capture a central component of real experiences: that our sensory input flows in a continuous way, and we naturally make choices about segmenting this experience into individual events that can be understood and remembered as units. Combining our two groups’ complementary expertises in intracranial neural recording and time-resolved analyses of multivariate electrophysiological data (Jacobs) and in naturalistic experiments using movies with meaningful event boundaries (Baldassano), we propose a unique set of studies to uncover how streams of sensory experiences are segmented and stored by the brain as discrete memories. This project provides the basis for a long-term collaboration to elucidate the detailed human neural mechanisms related to memory impairment in real-world settings.

Joshua Jacobs is an Associate Professor of Biomedical Engineering, whose research interests include human memory, direct brain recordings, stimulation, electrophysiology, and neuronal oscillations. Professor Jacobs and his laboratory examine the neural basis of human spatial navigation and spatial memory. This work is performed by conducting direct brain recordings from epilepsy patients that have electrodes implanted surgically in deep brain structures. Via these recordings, he and his team identify neural patterns that reveal how the brain represents memory for spatial locations and maps. Understanding this system is important not only for explaining how humans navigate, but also because it will elucidate how the brain supports various types of memory processes and suggest treatments for disorders such as Alzheimer’s Disease. The lab performs this work in close collaboration with neurosurgeons and neurologists at several hospitals, including Columbia University Medical Center, University of Pennsylvania, Thomas Jefferson University, Emory University, and University of Texas. There are several broader goals of this work. First, the team is interested in comparing the neural representation of space between humans and animals to identify common and distinctive aspects of spatial coding between species. Second, they test whether the neural coding of location during movement is similar to the brain patterns used to encode memories. Third, they engage in translational research to develop brain stimulation protocols for enhancing spatial memory to help people who experience cognitive impairment due to aging or disease. Professor Jacobs received Bachelor’s and Master’s degrees from the Massachusetts Institute of Technology in computer science in 2001 and 2002. In 2008, he received his PhD in neuroscience from the University of Pennsylvania.

Christopher Baldassano is an Assistant Professor of Psychology at Columbia University Medical Center. His research interests include perception, memory, cognitive neuroscience, and event cognition. Professor Baldassano earned his PhD from Stanford University in 2015 using machine learning methods to study the human visual system. He did his postdoc at the Princeton Neuroscience Institute before joining the Columbia Psychology faculty. As soon as he came to Columbia, Professor Baldassano formed the Dynamic Perception and Memory Lab, which studies how people can understand and remember the complex world of our everyday lives. Through experiments using narratives, movies, and virtual reality, his team investigates how experiences are divided into events, summarized, associated, and recalled. Professor Baldassano’s current projects are specifically focused on how prior knowledge about the temporal and spatial structure of the world influences human’s construction of mental representations. Using neuroimaging tools (primarily functional MRI), they can build models of how neural representations vary across stimuli and across people. His team employs both hypothesis-driven and data-driven approaches based on approaches from modern machine learning, which allow them to ask new kinds of questions about how brain regions respond to the world and interact with each other.

Pulmonary surfactant is a mixture of lipids and proteins that coats the alveoli and keeps them open, thus reducing the work of breathing and avoiding lung collapse. Surfactant is produced by alveolar epithelial type 2 (AT2) cells. Defects in the genes encoding the surfactant proteins can result in insufficient surfactant production and can cause severe respiratory distress in full-term infants, or childhood interstitial lung disease (chILD) in older children accompanied by respiratory failure and fibrosis. ChILD affects up to 16.2 patients per 100,000 people and has a mortality as high as 35%. There is no specific treatment, except for lung transplantation, which is hampered by a severe shortage of donor organs, especially for children. There is an urgent need for early treatment of chILD before end-stage lung disease sets in and affects those children for the rest of their lives. One alternative approach may be cellular replacement therapy: replace dysfunctional AT2 cells with healthy AT2 cells. To partially remove dysfunctional AT2 cells, we have engineered a recombinant protein in which a protein normally recognized and internalized by AT2 cells, surfactant protein A (SPA), is bound to a fragment of diphtheria toxin (DT388), DT388-SPA, which induces apoptosis once inside the cell. Once dysfunctional AT2 cells are cleared out by DT388-SPA, we will introduce new healthy AT2 cells. We hypothesize that those healthy AT2 cells will engraft in the treated areas of the lung and promote lung repair and functionality. We will test our hypothesis in two aims. In Aim 1, we will characterize the effect of AT2 cell engraftment after treatment with DT388-SPA in slowing or reversing lung disease in a mouse model of chILD, the Sftpc-/- mouse, that lacks surfactant protein C. In Aim 2, we will explore the translational potential of DT388-SPA to humans, directly testing its effect on human AT2 cells from lung specimens ex vivo. This proposal relies on three major innovative components: 1) a therapeutic recombinant protein, DT388-SPA, that targets AT2 cells, 2) a mouse model of chiLD to test cellular replacement therapy 3) human lung specimens to directly test DT388-SPA on human AT2 cells. The proposed strategy has the potential to significantly improve the prognosis of chILD and allow those children to live their life to their fullest.

Gordana Vunjak-Novakovic is the University Professor and Mikati Foundation Professor of Biomedical Engineering. Her diverse team of engineers, clinicians, and scientists are developing innovative tissue engineering technologies for improving human health. Our Laboratory for Stem Cells and Tissue Engineering is interested in whole organ engineering for regenerative medicine, tissue models for biological research, and “organs-on-a-chip” platforms for disease modeling and drug development. To this end, her team directs the human cell differentiation and assembly into functional tissues using a “cell-instructive” approach based on tissue-specific scaffolds (providing templates for tissue formation) and advanced bioreactors (providing environmental control, molecular and physical signaling). Her work has been published in Nature, Cell, Nature Biotechnology, Nature Medicine, Nature Biomedical Engineering, Nature Communications, Nature Protocols, PNAS, Cell Stem Cell, Science Advances, and Science Translational Medicine, and is highly cited (h=126).

Professor Vunjak-Novakovic’s laboratory is a home to the national Tissue Engineering Resource Center funded by NIH to foster tissue engineering for medical impact, and are actively collaborating with colleagues at both campuses of Columbia University, nationwide, and around the world. Dr Vunjak-Novakovic is also part of CELL-MET, a multi-institutional National Science Foundation Engineering Research Center in Cellular Metamaterials (EEC-1647837). CELL-MET aims to grow functional and clinically significant heart tissue while simultaneously developing a talented and diverse workforce to tackle future challenges in synthetic tissues engineering. To translate their science into new therapeutic modalities, their lab has launched four biotech companies: epiBone (epibone.com), Tara(tarabiosystems.com), Xylyx Biosolutions (xylyxbio.com) and Immplacate (immplacatehealth.com) that are all based in New York City. Over the last 30 years, Vunjak-Novakovic has mentored over 150 trainees (postdocs, clinical fellows, MD/PhD and PhD students, junior faculty). She has a BS, MS, and PhD in chemical engineering from the University of Belgrade and specialized in biomedical engineering as a Fulbright Fellow at MIT. Professor Vunjak-Novakovic’s is a member of the Academia Europaea, Serbian Academy of Arts and Sciences, the National Academy of Engineering, the National Academy of Medicine, the National Academy of Inventors, and the American Academy of Arts and Sciences.

Nicolino Dorrello is an Assistant Professor of Psychology at Columbia University Medical Center. His research interests include perception, memory, cognitive neuroscience, and event cognition. Professor Baldassano earned his PhD from Stanford University in 2015 using machine learning methods to study the human visual system. He did his postdoc at the Princeton Neuroscience Institute before joining the Columbia Psychology faculty. As soon as he came to Columbia, Professor Baldassano formed the Dynamic Perception and Memory Lab, which studies how people can understand and remember the complex world of our everyday lives. Through experiments using narratives, movies, and virtual reality, his team investigates how experiences are divided into events, summarized, associated, and recalled. Professor Baldassano’s current projects are specifically focused on how prior knowledge about the temporal and spatial structure of the world influences human’s construction of mental representations. Using neuroimaging tools (primarily functional MRI), they can build models of how neural representations vary across stimuli and across people. His team employs both hypothesis-driven and data-driven approaches based on approaches from modern machine learning, which allow them to ask new kinds of questions about how brain regions respond to the world and interact with each other.

Patients suffering from symptomatic spinal deformity have a significantly reduced quality of life, and are unable to perform daily routine activities. Various surgical options exist to optimize functional outcomes while limiting morbidity; Proximal Junctional Kyphosis (PJK) is one of these complications following spinal deformity surgery that may require re-operation and have a serious impact on a patient’s postoperative course. Our data features fall into several categories - demographics, surgical variables, pre-operative variables (pre-op), immediate post-operative variables, follow-up postoperative variables, and descriptive variables regarding final classification (1 or 0). However, during inference time, we only have access to demographics, surgical, pre-op, and immediate post-op variables. Hence, we are not able to use 33% of the features we have available in a conventional machine learning paradigm. When it comes to high risk tasks - such as patient-care - it is imperative to maximize the number of features we have available, regardless of the number of features left after pre-processing. The goal of our project is to build machine learning models for prediction and counterfactual explanations that can inform clinical decision making for future patients. To address this issue, we incorporate privileged data into XGBoost. Through the LUPI paradigm we are able to incorporate the missing 33% of features. Through our case study, we demonstrate the usefulness of the LUPI paradigm in machine learning for healthcare applications. This pilot study will set the stage for translational steps to bring our models into clinical use with the potential to transform outcomes for patients suffering from debilitating spinal deformity who undergo surgery.

Ansaf Salleb-Aouissi is a Senior Lecturer in the Discipline of Computer Science in the Department of Computer Science. Dr. Salleb-Aouissi’s specific and recent research interest is interdisciplinary and consists in leveraging advanced machine learning methods and large amounts of data to study medical problems, such as premature birth and infantile colic. Salleb-Aouissi cares about education and works toward advancing research on online self-learning and building advanced tools for auto-grading, self-testing, and providing support to students in computer science and mathematics. She has published several peer-reviewed papers in top quality venues including JMLR, TPAMI, ECML, PKDD, COLT, IJCAI, ECAI and AISTAT. Dr. Salleb-Aouissi joined the Department of Computer Science as a lecturer in discipline in July 2015. She received her PhD in computer science from University of Orleans, France in 2003, after which she pursued her training as a postdoctoral fellow at INRIA, Rennes (France). She was appointed as an associate research scientist at the Columbia University’s Center for Computational Learning Systems in 2006 and served as an adjunct professor with the Computer Science Department and the Data Science Institute in 2014 and 2015.

Larry Lenke is the Co-Director of Och Spine, surgeon-in-chief at the Och Spine Hospital, and chief of spinal deformity surgery in the Department of Orthopedic Surgery at NewYork-Presbyterian/Columbia University Irving Medical Center. Dr. Lenke is one of the world's foremost leaders in spinal deformity surgery. His world-renowned practice is devoted exclusively to spinal deformity surgery with an emphasis on complex reconstructive surgery in both children and adults for the treatment of various spinal deformities such as scoliosis, kyphosis, flatback syndrome, and other major spinal imbalances, as well as spondylolisthesis. He is generally regarded as the premier spinal deformity surgeon in the world, having developed the classification system for Adolescent Idiopathic Scoliosis (AIS), to which his name is now attached. After receiving his undergraduate degree from the University of Notre Dame and his MD from Northwestern University Medical School, Dr. Lenke completed his internship and residency training in Orthopaedic Surgery at Barnes-Jewish Hospital/Washington University School of Medicine. While at Washington University, he also completed his fellowship training in pediatric and adult orthopedic spine surgery. Dr. Lenke has been listed in America's Top Doctors for the past 10 years and Best Doctors in America the past 15 years. Dr. Lenke was honored with the North American Spine Society’s 2013 Leon Wiltse Award for excellence in leadership and/or clinical research in spine care. Also in 2013, Dr. Lenke was listed in Orthopedics This Week as one of “The Top 28 Spine Surgeons in North America.” He served as president of the Scoliosis Research Society 2010-2011, the oldest and most prestigious spine society in the world; its single focus is the advancement of care in patients with spinal deformity. As a reflection of his preeminent surgical skills, he has hosted over 700 spinal surgeons from around the globe to observe his surgeries in the past 15 years. Dr. Lenke’s prolific academic career includes writing over 360 published peer-reviewed manuscripts, editing five textbooks on Spinal Surgery, writing more than 125 textbook chapters, chairing over 100 Spinal Surgery meetings and having been an invited Visiting Professor domestically and internationally more than 100 times.

Joseph Lombardi is an orthopedic spine surgeon specializing in the operative management of spinal disorders in adult and adolescent patients. His practice is focused on minimally invasive surgical treatments for both common and complex spinal conditions, including disc herniation, spinal stenosis, spondylolisthesis, and spinal deformity. Dr. Lombardi leverages the latest surgical technology – robotics, guided navigation, and advanced instrumentation – to provide patients with tailored surgical procedures, accelerated recovery times, and superior outcomes. After receiving his undergraduate and medical degrees from The George Washington University in Washington, DC, Dr. Lombardi went on to complete his post-doctoral residency training in orthopedic surgery at Columbia University Irving Medical Center where he served as chief resident. Dr. Lombardi remained at Columbia and the NewYork-Presbyterian Och Spine Hospital for fellowship training in advanced adult and pediatric spine surgery. In addition to his clinical practice, Dr. Lombardi is involved in training future surgeons, and is an active member of the academic orthopedic spine community. His research interests are focused on the application of new technologies to complex spinal conditions with the goal of improving patient outcomes. To date, Dr. Lombardi has published over 15 book chapters, authored over 30 peer-reviewed journal articles, and has received numerous academic awards, including: the Alpha Omega Alpha honors society, Julius S. Neviaser Award in Orthopedic Surgery, Harold M. Dick Award for Excellence in Orthopedic Surgery, and the Alexander Garcia Award for Excellence in Clinical Orthopedic Surgery.

This proposal combines robotics, movement science, and clinical rehabilitation to address a critical societal need to improve walking in children with cerebral palsy. The proposed robotic rehabilitation is designed to make these children more balanced, improve their propulsion and gait speed, reduce their falls risk, and improve their quality of life. This proposal would help to quantify changes in walking characteristics of children with CP when external forces and moments are applied to their pelvis using mTPAD during gait training. The intervention with mTPAD will be performed both in single session and multiple training sessions to provide a proof of concept before multi-center randomized trials can be performed. The central innovation in the design of mTPAD is that it is the first low-cost mobile robotic platform to characterize and retrain walking over ground. mTPAD can apply phase-appropriate controlled forces and moments on the pelvis based on the phase of the gait cycle predicted using artificial intelligence deep learning approaches. The scientific goals of this project are to (i) characterize walking functions of children with CP with specific patterns of external pelvic forces/moments, and (ii) investigate their potential to retrain walking in children with CP over multiple training sessions.

Sunil Agrawal is a Professor of Mechanical Engineering and Professor of Rehabilitation and Regenerative Medicine. Professor Agrawal has developed a highly visible interdisciplinary program in rehabilitation robotics involving faculty from School of Engineering and Applied Sciences and College of Physician and Surgeons at Columbia University. Neural disorders, such as stroke and Parkinson’s disease, limit the ability of humans to walk and perform activities of daily living. Pediatric disorders such as cerebral palsy, spina bifida, and Down’s syndrome delay the development of children and pose many functional limitations. Old age diminishes the sensory and motor systems. Through a range of pilot and clinical studies involving human subjects, Professor Agrawal has showed that novel training robots can help humans to relearn, restore, or improve functional movements. He has active collaborations with faculty in the departments of Neurology, Rehabilitation Medicine, Pediatric Orthopedics, Otolaryngology, Geriatrics, and Psychiatry. A selected list of these ongoing studies are: (i) Perturbation training of the elderly using a Tethered Pelvic Assist Device (TPAD), (ii) Gait training of stroke patients with asymmetric forces, (iii) Balance training of children with cerebral palsy, (iv) Gait characterization of patients with vestibular disorders, (v) Balance Training of Parkinson patients, (vi) Novel neck braces for assistance and training of patients with head drop, (vii) Novel dynamic spine braces for patients with scoliosis. These studies are funded by grants from the National Science Foundation, National Institute of Health, Spinal Cord Injury Research Board, and others. Professor Agrawal received a BS in mechanical engineering from IIT, Kanpur (India) in 1984, a MS degree from Ohio State University in 1986, and a PhD degree in mechanical engineering from Stanford University, California, in 1990. He is a fellow of the American Society of Mechanical Engineers (ASME) and American Institute of Medical and Biological Engineering (AIMBE). He is an author of 450 research articles, 3 books, and 13 patents.

Hana Azizi is an Assistant Professor in Rehabilitation Medicine at Columbia University Irving Medical Center. She graduated from Shiraz University of Medical Sciences and completed her residency in Physical Medicine and Rehabilitation at Montefiore Medical Center. She then pursued a fellowship training in Pediatric Rehabilitation at Rusk Rehabilitation Institute/New York University Langone Health. She works with children and their families using an interdisciplinary approach to address the prevention, diagnosis, treatment, and management of congenital and childhood-onset physical impairments.

We will demonstrate a new paradigm for imaging in which implantable optoelectronic imagers record fluorescent signals from neurons in deep volumes of the mouse brain. In the proposed paradigm, the elements of a lensless microscopic functional imaging system, including microscale optical emitters to excite labeled neurons, electro-optical detectors to measure fluorescent signals, metasurfaces to shape the wavefronts of the excitation and fluorescent light, and electronic circuits, will be all fabricated on an ultranarrow shank with a cross-sectional dimension of 60 microns by 90 microns. The implantable imager emits rapid sequences of structured illumination patterns and records the resulting sequences of wavefront-shaped fluorescent emission from soma-labeled neurons. A blind-source-separation algorithm de-mixes and localizes individual fluorescent targets.

Nanfang Yu is an Associate Professor of Applied Physics and Applied Mathematics. Professor Yu studies the interaction between light and structured active materials at the nanometer scale and builds novel devices including lasers, detectors, and active components for controlling light. He has found that nanostructured materials usually turn out to be superb solutions for the generation, control, and detection of infrared waves with long wavelengths. For example, quantum cascade lasers, the most popular coherent infrared light sources, are made of semiconductor superlattices, which comprise a stack of nanometer-thick semiconductors, the thinnest layer being only a few atoms thick. Professor Yu and his lab work to create a new class of flat optical components, the thicknesses of which are just one-thousandth of a human hair and yet are able to do all the jobs conventional, bulky optical components can do. These flat optical devices are made of a two-dimensional array of nano optical scatters, which can individually and abruptly change the amplitude, phase and/or polarization of the scattered light. Together, the array of scatterers can dynamically mold optical wavefronts into arbitrary shapes with ultra-high speed. To build better infrared detectors and communication systems, Professor Yu probes the clever strategies developed by infrared-sensing organisms over their long evolutionary history. He is conducting interdisciplinary research with biologists and ecologists, exploring ways in which certain insects are able to detect broadband thermal radiation with extremely high sensitivity and spatial accuracy, or detect infrared “fingerprint” emissions from chemicals with high specificity. These studies will have a broad impact on novel heat-sensing devices and chemical detection systems. Professor Yu received a B.S. in Electrical Engineering from Peking University in 2004 and a Ph.D. in Engineering Sciences from School of Engineering and Applied Sciences at Harvard University in 2009. Yu joined Columbia University in 2013 after completing his doctoral and postdoctoral work at Harvard University. At Harvard, Professor Yu worked extensively on plasmonics, metamaterials, and mid-infrared and terahertz semiconductor lasers. He is a contributing member of a number of professional societies including the Optical Society of America, the IEEE Photonics Society, the American Physical Society, and the Materials Research Society.

Kenneth Shepard is the Lau Family Professor of Electrical Engineering and Professor of Biomedical Engineering. His research interests include integrated circuits and systems, systems biology and neuroengineering, smart electric energy, and computer engineering and computer systems. Kenneth L. Shepard received the BSE degree from Princeton University, Princeton, NJ, in 1987 and the MS and PhD degrees in electrical engineering from Stanford University, Stanford, CA, in 1988 and 1992, respectively. From 1992 to 1997, he was a Research Staff Member and Manager with the VLSI Design Department, IBM T. J. Watson Research Center, Yorktown Heights, NY, where he was responsible for the design methodology for IBM’s G4 S/390 microprocessors. Since 1997, he has been with Columbia University, New York, where he is now Professor of Electrical Engineering and Biomedical Engineering. He also was Chief Technology Officer of CadMOS Design Technology, San Jose, CA, until its acquisition by Cadence Design Systems in 2001. His current research interests include power electronics, carbon-based devices and circuits, and CMOS bioelectronics. Dr. Shepard was Technical Program Chair and General Chair for the 2002 and 2003 International Conference on Computer Design, respectively. He has served on the Program Committees for IEDM, ISSCC, VLSI Symposium, ICCAD, DAC, ISCAS, ISQED, GLS-VLSI, TAU, and ICCD. He received the Fannie and John Hertz Foundation Doctoral Thesis Prize in 1992, a National Science Foundation CAREER Award in 1998, and the 1999 Distinguished Faculty Teaching Award from the Columbia Engineering School Alumni Association. In 2014, he was named Lau Family Professor of Electrical Engineering. He has been an Associate Editor of IEEE Transactions on Very Large-Scale Integration (VLSI) Systems and is currently an Associate Editor for the IEEE Journal of Solid-State Circuits and IEEE Transactions on Biomedical Circuits and Systems. He is a Fellow of the IEEE.