Photo Credit: Greater Louisville Medical Society

By Randall C. Wetzel

The ethics of Science regards the search for the truth as one of the highest duties of man.

-Edwin Grant Conklin [1]

Decades into the digital revolution with our digital smartphones and our digital music and our increasingly digital patients, what are we doing with all these data?  If, as declared by the Financial Times: “We are Data” [2], then our patients are also data and in a big data way.  With apologies to Samuel Coleridge Taylor, healthcare may be more and more characterized by this rhyme of the modern clinician:

Data, Data everywhere,

And all the alarms did blink;

Data, data everywhere,

Nor any chance to think….

We may ask how we avoid drowning in this wonderful data deluge, but the more imperative question is how do we learn from it about and for our patients?  Further questions include: Should we be learning from it? How do we learn from it? What can we learn from it? Do we use these data from the myriad natural experiments that are performed in and on our patients every day to better care for our next patient? Can we be faithful to the ethics of science and discover ‘truth’ here?

The ethical principle of beneficence requires us to do the best for our patients.  This requires us to be constant learners. Learning from our patients is a core ethical principle of medicine and this has been so since the time of Hippocrates who exhorted us to learn from our patients by carefully collecting and recording the evidence about our patients and their illnesses [3]. Hippocrates declared that medicine should depend on detailed observation, experience and reason to establish prognosis, diagnosis and treatment. He insisted on the importance of meticulous observations and clinical documentation foreshadowing the use of our EMR databases. Hippocrates himself made considered and careful notes of patients’ symptoms, describing the importance of fever, the pulse, pain and its location and character, bodily excretions and the patient’s complexion. He included family and social history and environment factors to comprehensively understand the patient’s condition. This initiated modern medicine – yet we continue to fail to learn from observations about our patients and their interactions with their care providers. As Microsoft’s Eric Horvitz stated [4]:

ethical artificial intelligence

Photo Credit: Raed Mansour

“Nearly 2500 years ago, Hippocrates kicked off a revolution in healthcare by calling for the careful collection and recording of evidence about patients and their illnesses. This call, which first introduced the goal of sharing data among physicians to provide the best care possible for patients, established a foundation for the evolution of modern healthcare. Although 25 centuries have passed since Hippocrates’ call, we have not yet attained the dream of true evidence-based healthcare. Large quantities of data about wellness and illness continue to be dropped on the floor, rather than collected and harnessed to optimize the provision of care. We are simply not yet doing the best that we can.”

Further, Hippocrates belonged to a school of medical thought which emphasized prognosis over diagnosis. In his treatise, ‘On Forecasting Diseases’ Hippocrates stated [3]: “I believe that it is an excellent thing for a physician to practice forecasting. He will carry out the treatment best if he knows beforehand from the present symptoms what will take place later.” This implies Hippocratic physicians should predict outcomes based on observed data and probabilities. As we are increasingly learning, prognosis can be achieved by applying machine learning to medical data to predict outcomes as numerous warning algorithms, severity of illness scores and recent publications demonstrate [5, 6, 7].

It seems reasonable to imagine that Hippocratic ethics would strongly advocate for the application of the power of artificial intelligence to prognosticate, to use predictive analytics to forecast our patients’ conditions.

As for using data science to learn from the data, the New York Times asked in 2014 [8], “Can big data tell us what clinical trials don’t?” Medicine is inherently experimental because therapies rely on experimental results, and every intervention is, in one sense, an experiment in that the outcome is not certain.

We pride ourselves in thinking what we do has a randomized, double blind, well-controlled, hypothesis testing clinical trial somewhere behind it, even though we know we are deluding ourselves. The old hypothesis driven, RCT based science is increasingly hard to do, expensive and so restricted in external application that we must find another way [9].

Taking advantage of already collected data and learning from it provides an empirical basis for understanding what will happen to our patients.

Historically, observational empiricism was limited by shortcomings in human memory and computational power to make sense of myriad observations from natural experiments. It was subsequently eclipsed by experimental empiricism in the 17th and 18th centuries due to the tremendous experimental power to absolutely refute hypotheses. In our practices we have already performed thousands of experiments but thrown away the data!

With the advent of modern computational power and storage capabilities, an empirical approach based on the observed evidence and analyzed by machine learning provides a new scientific method – a paradigm shift for medical research.

We must take the opportunity to learn from every heartbeat, every breath, every lab value, and every bit of data we can capture. We owe it to our patients – it is our ethical responsibility.


The ethical imperative to learn from our patients’ data implies two further necessities: first getting the data [10] and second, assuring that the data are good.

We must stop wasting the data generated during our care processes, from the waveforms from our monitors to the data in our EMRs. Additionally, we must share the data to further the advances in analytic processes [].

Data ethics require that we assure the data are good, of sufficient quality for the task and of suitable analytic quality to prevent the “garbage in, garbage out” phenomenon. This is not so simple.

Medical data is unique, complex, massive and special [11]. Understanding the complexity of medical data is Informatics Research. Learning from the data is fundamentally different than capturing the data and presenting it at the bedside as we do with our EMRs. Organizing medical data in a defined open structure with defined semantic architecture, a common ontology and ready interpretability and access is also necessary [12].

But doing anything less means failing in a 2500-year-old medical ethics directive to learn from our patients. Failure to do so is unethical.

We must use the coolest new data science and search for truth in the best, ancient traditions of medicine or, like the ancient mariner, our failure to learn from our patients’ data may become the albatross around our necks.


  1. Conklin EG. From an address to the American Association for the Advancement of Science, Indianapolis 27 Dec 1937. In ‘Science and Ethics’. Science 66:2244, p. 602, 1937.
  2. We are Data. Review in The Financial Times. London and New York. July 19th, 2015.
  3. Parker Steve. Hippocrates and Greek Medicine. in Kill or Cure an Illustrated History of Medicine.  Angela Wilkes senior ed. pp 30-9. Dorling Kindersley NY, NY
  4. Eric Horvitz. From Data to Predictions and Decisions: Enabling Evidence-Based Healthcare. from Series on Data Analytics, Computing Community Consortium September 18, 2010. Downloaded on 3/12/2018 from:
  5. Vincent JLMoreno R. Clinical review: scoring systems in the critically ill. Crit Care.2010;14(2):207. doi: 10.1186/cc8204. Epub 2010 Mar 26.
  6. Benjamin M. Marlin, David C. Kale, Robinder G. Khemani, and Randall C. Wetzel. 2012. Unsupervised pattern discovery in electronic health care data using probabilistic clustering models. In Proceedings of the 2nd ACM SIGHIT International Health Informatics Symposium(IHI ’12). ACM, New York, NY, USA, 389-398. DOI=10.1145/2110363.2110408
  7. Williams J, Ghosh D, Wetzel RC. Applying Machine Learning to Pediatric Critical Care Data. In press.  PCCM
  8. Greenwood V. Can Big Data Tell Us What Clinical Trials Don’t? New York Times Oct 3 2014.
  9. Lauer MS, D’Agostino RB Sr. The randomized registry trial–the next disruptive technology in clinical research? N Engl J Med. 2013 Oct 24;369(17):1579-81. doi: 10.1056/NEJMp1310102. Epub 2013 Aug 31.
  10. Wetzel RC. First Get the Data, Then do the Science. In press. PCCM
  11. Ciosa KJ Moore GW. Uniqueness of medical data mining, Artificial Intelligence in Medicine 26 (2002) 1–24.
  12. J. Crichton, C. A. Mattmann, A. F. Hart, D. Kale, R. G. Khemani, P. Ross, S. Rubin, P. Veeravatanayothin, A. Braverman, C. Goodale, R. C. Wetzel, “An informatics architecture for the Virtual Pediatric Intensive Care Unit,” Computer-Based Medical Systems, IEEE Symposium on, pp. 1-6, 2011 24th International Symposium on Computer-Based Medical Systems, 2011.



ethical artificial intelligenceBy Randall C. Wetzel, Director, The Laura P. and Leland K. Whittier Virtual Pediatric Intensive Care Unit, Los Angeles CA

Dr. Randall Wetzel is the Chairman of the Department of Anesthesiology Critical Care Medicine at Children’s Hospital Los Angeles, Professor (Tenured) of Anesthesiology and Pediatrics, Keck School of Medicine, University of Southern California, and the Director of The Laura P. and Leland K. Whittier Virtual Pediatric Intensive Care Unit (VPICU) at Children’s Hospital Los Angeles. Prior to coming to CHLA, Dr. Wetzel held various faculty appointments in Anesthesiology and Critical Care Medicine at Johns Hopkins Hospital and University from 1981-1997.

After completing his undergraduate studies at Durham University in England, he went on to receive his medical degree from the Medical College of St. Bartholomew’s Hospital, London University, England. He moved stateside for his pediatric residency, then fellowship training in Critical Care and Pediatric Cardiology at Rainbow Babies and Children’s Hospital at University Hospitals in Cleveland, Ohio. He was subsequently a fellow in Pediatric Critical Care and Resident/Chief Resident in Anesthesiology/Critical Care Medicine at Johns Hopkins Hospital in Baltimore. Dr. Wetzel returned to England for a fellowship in Pediatric Anesthesia and Critical Care Medicine at the Hospital for Sick Children and Guy’s Hospital in London. In addition, he earned a Master of Science in Business (Medicine) from Johns Hopkins University (1997). He moved to Los Angeles in 1997.

His current interests in medical informatics led him to develop the VPICU; a Southern California Critical Care Telemedicine Network; he is CEO and Chair of VPS, LLC a Quality Improvement and Data management company with over 120 hospital clients; and he leads a complex big data research team developing data extraction, management, architecture and advanced analytic techniques for highly granular Pediatric Critical Care Data. He has received over $18 million in federal and private foundation grant funding for these projects. He has served as a board member for The American Board of Pediatrics (2000-2003); Editor-in-Chief, the Society for Pediatric Anesthesia (1988-1995). He has authored 94 peer-reviewed articles and 45 book chapters, he is editor of Roger’s Critical Care Medicine and Critical Heart Disease in Infants and Children; and has presented over 100 abstracts at various symposia.

Dr. Wetzel and his wife Ann have been married 34 years and currently reside in La Cañada, California. They have four wonderful daughters.