Portal de Periódicos da CAPES

Commentary: The end of one journey is the beginning of the next

2020; Elsevier BV; Volume: 162; Issue: 4 Linguagem: Inglês

10.1016/j.jtcvs.2020.06.091

1097-685X

Nicolas Zhou, Mara B. Antonoff,

Artificial Intelligence in Healthcare and Education

Central MessageClinical and administrative databases in cardiothoracic surgery each have strengths and limitations; their complementary roles highlight the need for data science and advanced data analytics.See Article page 1166. Clinical and administrative databases in cardiothoracic surgery each have strengths and limitations; their complementary roles highlight the need for data science and advanced data analytics. See Article page 1166. Data function as the new oil in the digital world. Health care may be one of the last industries to fully digitize, but we have long used data as part of the scientific method to make decisions for our patients. Now that we are consistently collecting data on all of our patients and analyzing them on a population level, do we have the ability to answer all clinical questions with the power of large databases? In this issue of the Journal, Kurlansky1Kurlansky P. The rocky exhilarating journey from data to wisdom.J Thorac Cardiovasc Surg. 2021; 162: 1166-1169Abstract Full Text Full Text PDF Scopus (3) Google Scholar expertly tells the tale of 2 databases. Developed out of the necessities to evaluate quality and cost of care, clinical and administrative databases are used to generate data analytics for risk models in cardiac surgery, as well as to evaluate reimbursement decisions as the cost of health care continues to expand. Administrative databases excel in modeling cost but are inadequate for assessing performance across patient risks, whereas clinical databases focus on patient outcomes. Numerous publications have resulted from such databases to guide our treatment algorithms. However, even the most robust clinical databases cannot capture every possible variable that may potentially lead to differences in outcomes across time periods. The limitations of these databases, as well as the logistics of maintaining an ever-expanding list of variables, lead to problems of scale and lack of information detailed enough for nuanced analyses. Although efforts to bridge the gap have been made by combing both types of databases,1Kurlansky P. The rocky exhilarating journey from data to wisdom.J Thorac Cardiovasc Surg. 2021; 162: 1166-1169Abstract Full Text Full Text PDF Scopus (3) Google Scholar there is still much to be desired with the current infrastructure, considering the incoming sources of Big Data that are high volume, high velocity, and high variety. Data science and use of advanced data analytics are sorely needed to provide the necessary infrastructure to generate the type of research needed to address specific questions and for us to practice precision medicine. Trained data scientists not only work to maintain Big Data; they also develop artificial intelligence (AI) tools to perform tedious tasks that would take thousands of hours of manual chart review to perform. In our own experience, we have worked collaboratively with data scientists to extract data across multiple data silos, as well as use natural language processing to build patient cohorts of interest.2Zhou N. Corsini E.M. Jin S. Barbosa G.R. Kell T. Antonoff M.H. et al.Advanced data analytics for clinical research part I: what are the tools?.Innovations. 2020; 15: 114-119Crossref PubMed Scopus (9) Google Scholar,3Zhou N. Corsini E.M. Jin S. Barbosa G.R. Kell T. Antonoff M.H. et al.Advanced data analytics for clinical research part II: application to cardiothoracic surgery.Innovations (Phila). 2020; 15: 155-162Crossref PubMed Scopus (4) Google Scholar These are just the beginning steps of using an arsenal of AI tools to address research questions that would have been prohibitively labor intensive. In the future, the use of other tools such as optical character recognition,2Zhou N. Corsini E.M. Jin S. Barbosa G.R. Kell T. Antonoff M.H. et al.Advanced data analytics for clinical research part I: what are the tools?.Innovations. 2020; 15: 114-119Crossref PubMed Scopus (9) Google Scholar convolutional neural networks,4Soffer S. Ben-Cohen A. Shimon O. Amitai M.M. Greenspan H. Klang E. Convolutional neural networks for radiologic images: a radiologist's guide.Radiology. 2019; 290: 590-606Crossref PubMed Scopus (150) Google Scholar and many other forms of AI may serve to improve clinical and administrative data in accuracy, detail, and completeness. Lastly, health care data are no longer isolated to hospital systems alone. Technology companies have advanced into the health care space, and they are collecting data on patients that neither clinical nor administrative databases could achieve—continuously, automatically, and in the comfort of patients' homes. The Apple Heart Study garnered much attention for its feasibility of using a wearable device in detecting atrial fibrillation in a transient general population.5Perez M.V. Mahaffey K.W. Hedlin H. Rumsfeld J.S. Garcia A. Ferris T. et al.Large-scale assessment of a smartwatch to identify atrial fibrillation.N Engl J Med. 2019; 381: 1909-1917Crossref PubMed Scopus (412) Google Scholar This study recruited more than 400,000 patients in a matter of 8 months. The speed and scale of this study was a sneak peek into a new territory of collaborative research with health care and technology companies. Cardiothoracic surgeons should take notice of the rapid development of noninvasive devices and their potential to change practice patterns. We urge our readers to have a basic understanding of data science and to be ready to adopt new technology into our patient care. The rocky exhilarating journey from data to wisdomThe Journal of Thoracic and Cardiovascular SurgeryVol. 162Issue 4PreviewEvery word or concept, clear as it may seem to be, has only a limited range of applicability. —Werner Heisenberg Administrative versus clinical databases: which is better? Yes. Or perhaps, neither. And why is it that we care? Full-Text PDF