Dr. Leo Anthony G Celi

Principal Research Scientist
Assistant Professor of Medicine, Beth Israel Deaconess Medical Center (BIDMC)

Primary DLC

Institute for Medical Engineering and Science

MIT Room: E25-505

Areas of Interest and Expertise

Healthcare: Value-Based, Triple Aim, Future Models

Research Summary

(*) Secondary Analysis of Electronic Health Records: A 2012 report from the Institute of Medicine stated that only 10-20% of clinical decisions are based on experimental data. In addition, the reliability of research is coming under increasing scrutiny. The National Institute of Health invests $30.9 billion annually in medical research. But despite the peer review process, there is a fundamental problem with the reliability of a disturbing amount of scientific, particularly biomedical, research. The problem is basically two-fold and lies in the unreliability of what does get published and the inability of other researchers to know what is not getting published. False positive research results are successfully published with alarming frequency. Replication is intrinsically difficult for a variety of reasons. And when re-examination does occur, irreproducibility is surprisingly rampant.

The good news is that solutions are available. But it will take a massive paradigm shift. The entire biomedical research industry needs nothing less than a different legal/government/industry framework for supporting a new model of digital, open and transparent data-driven health care.

As clinical research director of the Laboratory of Computational Physiology, I bring together clinicians and data scientists to support research using data routinely collected in the ICU. The laboratory developed and maintains the Multi-parameter Intelligent Monitoring in Intensive Care (MIMIC) database. This public-access database holds clinical data from over 60,000 stays in the intensive care units (ICU) of the Beth Israel Deaconess Medical Center. It has been meticulously de-identified and is freely shared online with the research community.

The vision is for the development of a care system consisting of “clinical informatics without walls”, in which the creation of evidence and clinical decision support tools is initiated, updated, honed, and enhanced by crowd sourcing. In this collaborative medical culture, knowledge generation would become routine and fully integrated into the clinical workflow. This system would employ individual data to benefit the care of populations and population data to benefit the care of individuals.


(*) Global Health Informatics: Dr. Celi founded and currently leads Sana (sana.mit.edu), a cross-disciplinary organization at the Harvard-MIT Division of Health Science and Technology (HST). It consists of doctors, engineers, public health practitioners, informaticians, and social entrepreneurs, among others. At its core is an open-source mobile tele-health platform that allows for capture, transmission and archiving of complex medical data (e.g. images, videos, physiologic signals such as ECG, EEG and oto-acoustic emission responses), in addition to patient demographic and clinical information. Sana initiates the design process with the identification of a clinical need by committed international partners, which include universities, NGOs, governments, and private ventures, with strong emphasis on local stakeholders. Sana champions 3 core principles. Firstly, any technological solution should be initiated as a response to a clinical need identified by the partner. Secondly, the endeavor is accompanied by the concomitant creation of a sustainable culture of quality and safety. Lastly, there should be a strong focus on capacity building and collaboration throughout both the design and implementation processes. The group maintains that, while a culture of quality and safety is crucial during the design phase, pilot and implementation, it is most essential during the scale-up. The mobile health information system is regarded as a tool to facilitate this larger aim: This is achieved by technical support of longitudinal and continuous care as well as provision of a platform for communication, care coordination, and decision support. Mobile health also provides a tracking system to measure processes and outcomes, which ultimately is used to improve quality.

To accomplish Sana’s mission of capacity-building, Celi directs a course offered at HST on innovations in global health informatics since 2011 that focuses on the design, implementation, evaluation and scale-up of health information systems in resource-poor settings. The goal is to incubate innovations within its partner universities, while learning from and creating best practices in the implementation and scale-up of information and communication technology in healthcare. The partners are not merely sources of specifications and feedback, but active participants in the co-creation of systems design at the highest level. Sana believes geniuses abound in its partner countries, and these geniuses are more likely to develop sustainable and scalable solutions as they understand the problems better. The course, which is free of charge and available online, has had over 500 students from around the world, including Colombia, the Philippines, India, Tunisia, Brazil, Uganda and Taiwan.


(summary updated 7/2015)

Recent Work

  • Video
    July 2, 2020Conference Video Duration: 60:21

    2020 Leo Celi

    2020 Leo Celi
    September 14, 2016Conference Video Duration: 24:10

    Leo Celi - 2016-Digital-Health_Conf-videos

    Bridging the Gap: Towards machine learning that matters in healthcare

    The Agency for Healthcare Research and Quality was established in 1989 in response to an Institute of Medicine report that pointed out ?escalating healthcare costs, wide variations in medical practice patterns, and evidence that some health services are of little or no value?. More than 25 years later, there has been surprisingly little progress in these three areas. The interest in applying machine learning to clinical practice is increasing yet the practical application of these techniques has been less than desirable. There is a persistent gap between the clinicians required to understand the context of the data and the engineers who are critical to extracting useable information from the increasing amount of healthcare data that is being generated.

    2016 MIT Digital Health Conference