Published on May 22, 2020
1. May 2020 Applying Network Medicine methodologies to study and treat SARS-CoV-2 infected patients MNM COVID-19 Study
2. OVERVIEW WHO ARE WE? WHAT IS NETWORK MEDICINE? The International Network Medicine Foundation and Consortium represents 28 leading medical centers committed to improving global health and advancing the field of Network Medicine. Network Medicine combines principles and approaches from network sciences, systems biology, and human dynamics to understand the causes of human diseases and develop new treatments. WHAT’S THIS HAVE TO DO WITH COVID-19? We are raising $1M to launch a three-phase clinical study to apply Network Medicine methodologies to study and treat SARS-CoV-2 infected patients.
3. MULTINATIONAL NETWORK MEDICINE COVID-19 STUDY (MNM COVID-19 STUDY) The MNM COVID-19 study is spearheaded by researchers at Brigham and Women's Hospital, one of Harvard Medical School’s flagship teaching hospitals. $700M+ The Brigham oversees the second largest hospital-based research program in the world. Brigham Annual Research Budget The Channing Division of Network Medicine at the Brigham is renowned as the site of the Nurses' Health Study. Now in its third generation, the studies count more than 280K participants and have influenced health policy and clinical care standards around the world.
4. MULTINATIONAL NETWORK MEDICINE COVID-19 STUDY (MNM COVID-19 STUDY) The Brigham’s Channing Division of Network Medicine (CDNM) CDNM uses an integrated network-based, systems biology-driven approach to determine the causes of complex diseases, reclassify them, and develop new treatments and preventive strategies. CDNM applies novel network science and artificial intelligence capabilities to its vast clinical databases and research studies with genetic, clinical, and epidemiological information on hundreds of thousands of patients. CDNM has more than 80 Harvard Medical School (HMS) faculty and 42 fellows in addition to 160 non-faculty Brigham and Women’s Hospital (BWH) employees, who are primarily in research and administration. In fiscal year 2019, CDNM investigators received 54 new funding awards resulting in 173 active grants. CDNM’s FY18 total research expenditures represented 25% of the Department of Medicine annual budget.
5. Our co-partner, The Karolinska Institute in Sweden, whose Nobel Assembly decides the Nobel prize in Medicine and Physiology, is committed to lead the European branch of the study. MNM COVID-19 STUDY …and other distinguished INMC consortium members and affiliates will contribute additional scale, capabilities, and an international presence to the MNM COVID-19 Study.
6. Until a vaccine is developed and novel treatments are approved, repurposing existing drugs is our best chance to help infected patients and try to limit the social and economic devastation caused by this virus. 1) Which contributing factors are protective or detrimental to an individual's response to SARS-CoV-2 infection? and how do combinations of such factors explain the varied clinical outcomes observed which range from asymptomatic to severely ill requiring hospitalization? MNM COVID-19 STUDY KEY QUESTIONS 2) Which currently available drugs or combinations would be most effective when repurposed to treat specific patient groups?
7. THE MNM COVID-19 STUDY HAS THREE PHASES Phase One: Collect Data Phase Two: Analyze Data Phase Three: Publish Findings and Expand AccessIn the first phase, we’ll establish a multinational patient registry in a format conducive to analysis using state-of-the-art Network Science and Artificial Intelligence techniques. The registry will include demographic, genomic, and clinical data for symptomatic and asymptomatic SARS-CoV-2 infected patients.
8. THE MNM COVID-19 STUDY HAS THREE PHASES Phase One: Collect Data Phase Two: Analyze Data Phase Three: Publish Findings and Expand AccessThen we’ll analyze the data to answer two fundamental questions: 1) Which contributing factors are protective or detrimental to an individual's response to SARS-CoV-2 infection? And how do combinations of such factors explain the varied clinical outcomes? 2) Which currently available drugs or combinations would be most effective when repurposed to treat specific patient groups defined by the analysis (eg. asymptomatic, mild, or significantly ill)?
9. THE MNM COVID-19 STUDY HAS THREE PHASES Phase One: Collect Data Phase Two: Analyze Data Phase Three: Publish Findings and Expand AccessIn the third phase, we’ll update and publish clinical care standards including the drug repurposing program. We’ll also expand access to our registry and data analysis capabilities through research collaborations with public and private institutions to influence health policy and answer other important questions as we learn more about this complex disease.
10. Biomedical data is distributed over countless small and large repositories around the world. So regulatory and technical barriers prevent collaborative use of advanced techniques such as machine learning to analyze these complex data sets comprised of clinical data, -omics data, and other metadata. To understand the varied disease manifestations of COVID-19, we need a way to overcome these barriers. HOW DO YOU ANALYZE DISTRIBUTED CLINICAL DATA WITHOUT COMPROMISING PATIENT PRIVACY? THE CHALLENGE Our Federated Analysis Data System (FADS) is built on a decentralized cloud computing infrastructure that does not require raw data to be transferred to a central site for analysis. THE SOLUTION
11. WHAT IS FADS? FADS uses federated learning to analyze medical data. Federated learning is an artificial intelligence (AI) technique that can analyze data stored at multiple locations, such as hospitals and outpatient clinics caring for patients with COVID-19, without the data ever leaving those locations. This means FADS enables researchers to analyze medical data at some of the best hospitals and outpatient clinics in the US, Europe, and the rest of the world in order to better understand the varied clinical outcomes observed in COVID-19 patients.
12. FADS IN A COVID-19 SETTING Step One: Define Common Data Model Representative Data in the COVID-19 CDM Patient Demographics Age, sex, race, socioeconomic, etc. Outpatient and Inpatient Medical Record Past and current medical conditions, interventions, medications, vaccinations, lab and imaging results, etc. COVID-19 History Exposure history, timing, location, and circumstances Progression of patient symptoms, clinical observations, lab and imaging results Dosage and timing of medications Therapeutic interventions needed for treatment Observed results of the medications and therapeutic interventions Outpatient outcomes and if transitioned to inpatient setting Evidence of any long-term complications The Common Data Model (CDM) enables researchers to analyze disparate observational databases. The concept behind this approach is to transform data contained within those databases into a common format with common terminologies, vocabularies, and coding schemes.
13. FADS IN A COVID-19 SETTING Step Two: Analyze the Data and Deploy Predictive Models Researchers develop a master set of predictive models based on historical COVID-19 patient data to predict clinical outcomes and identify drugs that can be repurposed to treat specific patient groups defined by the analysis. Federated learning techniques enable researchers to deploy copies of the models at each hospital and outpatient clinic participating in the study while maintaining master predictive models that don’t require access to underlying patient data. Master Set of Predictive Models Copies of Master Predictive Models Deployed Copies of Predictive Models at Hospitals and Outpatient Clinics
14. FADS IN A COVID-19 SETTING Step Three: Master Set of Predictive Models Improves Over Time The master set of predictive models improves each time a deployed model returns an updated representation of clinical data. This is an iterative process that accelerates as more hospitals and outpatient clinics join the study. Deployed Models Learn from Patient Data at Hospitals and Outpatient Clinics Deployed Models are Sent Back Periodically to Update Master Set of Predictive Models More Deployed Models = Better Predictions from Master Set of Models
15. 60%20% 10% 10% Setup infrastructure and data collection at 12 sites Central operations for IT infrastructure and data collection Federated Analysis Data System (FADS) Research coordination and operations Our initial fundraising target is $1M, which enables us to establish the multinational patient registry and launch the three phase COVID-19 clinical study. $1M FUNDRAISING TARGET USE OF PROCEEDS
16. MNM COVID-19 STUDY EXECUTIVE COMMITTEE Joseph Loscalzo, MD, PhD Chairman BOD and Scientific Committee of the Foundation and Head of Consortium and Hersey Professor of the Theory and Practice of Medicine, Soma Weiss, MD, Distinguished Chair in Medicine: Harvard Medical School. Chairman, Department of Medicine and Physician-in-Chief: Brigham and Women’s Hospital, USA Jan Baumbach, PhD Professor and Chair of Experimental Bioinformatic, Director of Information Technology (ITW), TUM School of Life Sciences (WZW), Technical University of Munich (TUM), Germany Eugenio Gaudio, MD Rettore of the Università, Professor of Medicine, Sapienza Università di Roma, Italy Jean-Luc Balligand, MD, PhD President of Experimental and Clinical Research UC Louvain, Professor of cardiovascular physiology and pharmacology, Physician at Cliniques Universitaire Saint-luc, Universite` Catholique de Louvain (UC Louvain), Belgium Albert-Laszlo Barabasi, PhD Robert Gray Dodge Professor of Network Science, Distinguished University Professor, Director of Center for Complex Network Research, Dept. Of Physics, Northeastern University, Channing Division of Network Medicine, Brigham and Women’s Hospital, Harvard Medical School, USA, Director Network and Data Science, Central European University, Hungary Sebastiano Filetti, MD Professor of Medicine, Delegate for "International Relations Health Care”, Sapienza Università di Roma, Italy Paolo Parini, MD, PhD Professor, Senior Consultant, Director of Research & Development, Education and Innovation, Head of Inflammation and Infection, Karolinska University Hospital, Department of Medicine Karolinska Institutet at Huddinge University Hospital, Sweden Enrico Petrillo, MD Executive Director of the International Network Medicine Foundation and Consortium, Advisor to Chairman, Department of Medicine and Physician-in-Chief for International Research, Lead Investigator, Division of General Internal Medicine, Brigham and Women’s Hospital, Lecturer, Harvard Medical School, USA
17. KEY TAKEAWAYS COVID-19 BUT NOT ONLY COVID-19 The COVID-19 patient registry and Federated Analysis Data System (FADS) enable unprecedented analytic capabilities in a clinical setting and will lead to publications, public health policies, and updated clinical care standards to help COVID-19 patients. IT’S TIME TO ACT These new capabilities will have a lasting effect on the medical community. Federated learning can untap the full potential of Network Medicine leaving us better equipped to address global health issues and better prepared for the next pandemic. We would greatly appreciate if you’d consider a personal contribution or corporate sponsorship to help us launch the MNM COVID-19 Study, establish the patient registry, and begin onboarding hospitals and outpatient clinics to FADS so we can do our best to bring an end to this crisis.
18. Thank You *Please continue to the appendix for additional MNM COVID-19 Study Resources and Network Medicine References Enrico Petrillo, MD Brigham and Women’s Hospital [email protected] MNM COVID-19 STUDY Paolo Parini, MD, PhD Karolinska Institute and University Hospital [email protected]
19. INMC and Foundation Mission Statement, Network Medicine Applied to Combat COVID-19 This memo expands on why Network Medicine is an essential strategy for the fight against COVID-19. INMC Executive Committee Cover Letter, MNM COVID-19 Outpatient Study INMC Executive Committee Cover Letter, MNM COVID-19 Inpatient Study The INMC executive committee describes the methodology and goals of the proposed MNM COVID-19 outpatient and inpatient studies. Framework and Outline, MNM COVID-19 Outpatient Study Framework and Outline, MNM COVID-19 Inpatient Study These provide a framework and outline of the three phase MNM COVID-19 outpatient and inpatient studies. Key questions to be addressed by MNM COVID-19 Outpatient Study Key questions to be addressed by MNM COVID-19 Inpatient Study These describe the key questions to be addressed by each branch of the MNM COVID-19 study. Federated Analysis Data System (FADS) The multifaceted disease states of COVID-19, complex regulatory landscape, and advanced Network Medicine techniques require a novel approach to data storage and analysis. This memo describes the proposed Federated Analysis Data System that will manage the international patient registry we create during Phase One of the MNM COVID-19 study. Standardized Protocols and Common Data Model for Federated Analysis Data System (FADS) This describes the Standardized Protocols and Common Data Model required to manage and analyze complex clinical data collected from repositories around the world. Scientific Rationale for the MNM COVID-19 Study This references featured papers in peer reviewed journals and provides a scientific rationale for the proposed MNM COVID-19 study. *For more information about the International Network Medicine Consortium’s Foundation which is sponsoring the study please visit www.intnm.org APPENDIX MNM COVID-19 STUDY RESOURCES
20. Molecular networks in Network Medicine: Development and applications Edwin K. Silverman, Harald H. H. W. Schmidt, Eleni Anastasiadou, Lucia Altucci, Marco Angelini, Lina Badimon, Jean-Luc Balligand ,Giuditta Benincasa, Giovambattista Capasso, Federica Conte, Antonella Di Costanzo, Lorenzo Farina, Giulia Fiscon, Laurent Gatto, Michele Gentili, Joseph Loscalzo, Cinzia Marchese Claudio Napoli, Paola Paci, Manuela Petti, John Quackenbush, Paolo Tieri, Davide Viggiano, Gemma Vilahur, Kimberly Glass, Jan Baumbach Systems Biology and Medicine March 20, 2020 Machine Learning Characterization of COPD Subtypes: Insights From the COPDGene Study Castaldi PJ, Boueiz A, Yun J, Estepar RSJ, Ross JC, Washko G, Cho MH, Hersh CP, Kinney GL, Young KA, Regan EA, Lynch DA, Criner GJ, Dy JG, Rennard SI, Casaburi R, Make BJ, Crapo J, Silverman EK, Hokanson JE; COPDGene Investigators. Chest. 2019 Dec 28. pii: S0012-3692(19)34456-3. doi: 10.1016/j.chest.2019.11.03 A genome-wide positioning systems network algorithm for in silico drug repurposing Feixiong Cheng, Yuan Hou, Diane E. Handy, Ruisheng Wang, Yuzheng Zhao, Yi Yang, Jin Huang, David E. Hill, Marc Vidal, Charis Eng & Joseph Loscalzo Nature Communications volume 10, Article number: 3476 (2019) The exposome and health: Where chemistry meets biology Roel Vermeulen, Emma L. Schymanski, Albert-László Barabási, Gary W. Miller Science 24 Jan 2020: 367, 6476, 392-396 The unmapped chemical complexity of our diet Albert-László Barabási, Giulia Menichetti & Joseph Loscalzo Nature Food 1, 33-37 (2019) Network-based prediction of protein interactions István A. Kovács, Katja Luck Yang Wang, Carl Pollis, Sadie Schlabach, Wenting Bian, Dae-Kyum Kim, Nishka Kishore, Tong Hao, Michael A. Calderwood, Marc Vidal & Albert-László Barabási Nature Communications 10, Article number: 1240 (2019) Network-based prediction of drug combinations Feixiong Chen, István A. Kovács & Albert László Barabási Nature Communications 10, Article number: 1197 (2019) Network-based approach to prediction and population- based validation of in silico drug repurposing Feixiong Cheng, Rishi J. Desai, Diane E. Handy, Ruisheng Wang, Sebastian Schneeweiss, Albert-László Barabási & Joseph Loscalzo Nature Communications, volume 9, Article number: 2691 (2018) APPENDIX NETWORK MEDICINE REFERENCES
21. The exposome and health: Where chemistry meets biology Roel Vermeulen, Emma L. Schymanski, Albert-László Barabási, Gary W. Miller Science 24 Jan 2020: 367, 6476, 392-396 January 24, 2020 The unmapped chemical complexity of our diet Albert-László Barabási, Giulia Menichetti & Joseph Loscalzo Nature Food 1, 33-37 (2019) December 9, 2019 Nature’s reach: narrow work has broad impact Alexander J. Gates, Qing Ke, Onur Varol & Albert-László Barabási Nature 575, 32-34 (2019) November 7, 2019 Network-based prediction of drug combinations Feixiong Chen, István A. Kovács & Albert László Barabási Nature Communications 10, Article number: 1197 (2019) March 13, 2019 Taking Census of Physics Federico Battiston, Federico Musciotto, Dashun Wang, Albert-László Barabási, Michael Szell, and Roberta Sinatra Nature Reviews Physics 1, 89-97 (2019) January 8, 2019 A Structural Transition in Physical Networks Nima Dehmamy, Soodabeh Milanlouei & Albert-László Barabási Nature 563, pages676–680 (2018) The Fundamental Advantages of Temporal Networks A. Li, S. P. Cornelius, Y.-Y. Liu, L. Wang, A.-L. Barabasi Science 358:6366, 1042-1046 (2017). The Elegant Law that Governs Us All A.-L. Barabasi Science 357:6347 (2017) Network Science Albert-László Barabási Philosophical Transactions of The Royal Society 371, 1-3 (2013) APPENDIX NETWORK SCIENCES REFERENCES