Temporal pattern recognition in multiparameter ICU data

Intensive Care Unit (ICU) patients are physiologically fragile and require vigilant monitoring and support. The myriad of data gathered from biosensors and clinical information systems has created a challenge for clinicians to assimilate and interpret such large volumes of data. Physiologic measurements in the ICU are inherently noisy, multidimensional, and can readily fluctuate in response to therapeutic interventions as well as evolving pathophysiologic states. ICU patient monitoring systems may potentially improve the efficiency, accuracy and timeliness of clinical decision-making in intensive care. However, the aforementioned characteristics of ICU data can pose a significant signal processing and pattern recognition challenge---often leading to false and clinically irrelevant alarms. We have developed a temporal database of several thousand ICU patient records to facilitate research in advanced monitoring systems. The MIMIC-II database includes high-resolution physiologic waveforms such as ECG, blood pressures waveforms, vital sign trends, laboratory data, fluid balance, therapy profiles, and clinical progress notes over each patient's ICU stay. We quantitatively and qualitatively characterize the MIMIC-II database and include examples of clinical studies that can be supported by its unique attributes. We also introduce a novel algorithm for identifying "similar" temporal patterns that may illuminate hidden information in physiologic time series. The discovery of multi-parameter temporal patterns that are predictive of physiologic instability may aid clinicians in optimizing care. In this thesis, we introduce a novel temporal similarity metric based on a transformation of time series data into an intuitive symbolic representation. The symbolic transform is based on a wavelet decomposition to characterize time series dynamics at multiple time scales. The symbolic transformation allows us to utilize classical information retrieval algorithms based on a vector-space model. Our algorithm is capable of assessing the similarity between multi-dimensional time series and is computationally efficient. We utilized our algorithm to identify similar physiologic patterns in hemodynamic time series from ICU patients. The results of this thesis demonstrate that statistical similarities between different patient time series may have meaningful physiologic interpretations in the detection of impending hemodynamic deterioration. Thus, our framework may be of potential use in clinical decision-support systems. As a generalized time series similarity metric, the algorithms that are described have applications in several other domains as well. Thesis Supervisor: Roger G. Mark Title: Distinguished Professor of Health Sciences and Technology Professor of Electrical Engineering and Computer Science Acknowledgments When I first came to MIT, some people told me that it would be a challenging and daunting environment. There have certainly been occasions when times were difficult--but in truth---this experience has been an enjoyable, life-altering, and blissful chapter in my life. As I am writing these acknowledgements, I fully realize that I am a wealthy man. I am wealthy because I have a circle of friends and family that have showered me with their generosity, love, and happiness. Professor Roger Mark has been a mentor, friend, and role model. He taught me a great deal about physiology, clinical medicine, and physiologic signal processing. I feel truly blessed for having met such a wonderful human being. We could just as easily converse about medicine, engineering, religion, politics, and family. His level of compassion for others and passion for educating and mentoring students is unrivaled. I wish him and his family all the joy and blessings that they so richly deserve. Professor Peter Szolovits has been a wonderful research mentor and great teacher. He has injected insightful ideas when I needed them, and gently prodded me forward with his great sense of humor when I may have tended to stagnate. Dr. Bill Long has been a valuable member of my doctoral committee, and has provided me with a great deal of practical advice as I explored different aspects of my thesis. Dr. Danny Talmor and Dr. Atul Malhotra are great colleagues to work with at BIDMC. Their insightful comments were particularly helpful in the clinical studies we developed from the MIMIC-II database. I look forward to many years of collaboration with them. The people at LCP have made my studies at MIT extremely enjoyable. There are few words that can describe my admiration and feelings for Rama Mukkamala. I feel extremely fortunate to know somebody who is so hard-working, creative, passionate, loyal, and sincere. He really brought the word "fun" into LCP. When I start a ten minute phone call with him, it usually ends two hours later because we have so much to discuss. I treasure our close friendship and I am grateful for his support over the years. Wei Zong is a constant professional in all senses of the word. He is unfailing in always providing his colleagues a helping hand. It is rewarding to know somebody who is so talented---yet so humble. I very much enjoyed every minute spent with Wei at MIT, Philips Medical Systems, and the swimming pool! Thomas Heldt has been a great colleague over the years. As my officemate, he was an intelligent, principled, and hard-working peer. Outside of the lab, he has been a loyal and supportive friend. I could always count on him to make sure that I didn't take myself too seriously with his witty zingers. Omar Abdala is a great friend and colleague. I am very proud to know somebody who can combine his level of intellect with a passion for making our greater community a better place. Matt Oefinger was great to have in the lab as a friend and colleague. His great sense of humor, positive outlook, and good nature were an integral part of LCP. Admittedly, I have yet to get over my loss to him in the LCP pizza-eating contest. George Moody has provided me with so much help over the years. As we developed the MIMIC-II database, his experience was invaluable in moving us in the right direction. His passion for facilitating research has helped to sustain LCP over the years. Andrew Reisner's wit, insight, and great sense of humor over the years were a welcome addition to LCP. I am grateful for the many times he taught me some very practical clinical medicine and pearls of wisdom about the dynamics of the medical profession. Sherman is not only a virtuoso of the violin, but also a talented student and kind friend who is great to work with. Professor George Verghese has made extremely valuable contributions to the MIMIC-II project. It is a pleasure to work with such a friendly and insightful individual. Dr. Joseph Frassica has been a valuable mentor for many years. His understanding of ICU medicine and the patient monitoring industry is amazing. Gari Clifford is an English gentleman in every respect. He is helpful to people in the lab and has a great sense of humor. Mauro, Zaid, Li-wei, Isaac, Christine Leu, Anton, Zaid, Dewang, Brian Janz, and Benjamin Moody were all great contributors to LCP and wonderful people to work with. Within the HST community, I have met many wonderful people. Ali Shoeb has been a great friend over the years. His exuberance for learning and positive attitude with respect to every activity he engages in serves as a great example of the model student. His energetic personality makes every conversation with him enjoyable and truly memorable. I first knew Lucia Madariaga as my team member in HST anatomy, and now I know her as my friend. Through the many enjoyable walks between Harvard and MIT, I came to appreciate and respect her spirituality, intelligence, and genuine kindness. I have known Nick Houstis since high school in Indiana. His great sense of humor, loyalty to his friends, and absolute brilliance make him one of the most amazing people I have ever met. Ray "L" Chan has been a great friend over the years. His unforgettable laughter and good nature always brightened my days. His positive outlook on life is an example I strive for. Quite possibly, Ying Zhang takes the title as the world's most considerate person. I remember when we had offices next to one another at Philips, she would bring me a glass of water when she heard the slightest cough out of me. You would think that somebody as kind as her is too good to be true---but she is for real! I have had the pleasure of knowing some truly gifted and wonderful people in EECS over the years. It has been an honor to know Professor Paul Gray at MIT. I enjoyed being his TA in circuit class. He was always available to render his wealth of invaluable advice to me when I needed it. It was also a pleasure serve as a TA under Professors Art Smith, Jacob White, and Jeff Lang. In the many environments I have worked in, there were always capable people making sure I didn't stray too far off the path. Marilyn Pierce of EECS at MIT was always there for any questions I had about registration, graduation, and institute policies. Patty and Linda have been so helpful over the years at the HST office. At LCP, Ken has been extremely helpful anytime I needed his expertise in document processing or filling out obscure forms. Karen at Philips was always available to help me in every thing I did when all hope seemed lost! Professor Rick Mitchell at HST has always been a great leader and advisor for me and countless other students. The first month I came to MIT as a na'ive twenty one year-old kid from Indiana, I felt out of place and longed for the comforts of home and my family. While my dorm was next to a nuclear reactor, by the grace of God, my next door neighbor happened to be Gassan AlKibsi. Since I have known him, he has been my truest friend and brother. He has been an unfailing source of support, comfort, inspiration, and optimism. Whenever I feel challenged by a situation professio