Big datasets, combined with advancements in artificial intelligence (AI) and machine learning (ML), are transforming research, education, and the delivery of next-generation clinical care.

At NYU Langone’s Health Tech Datathon, we bring together leaders in clinical and data science to tackle complex clinical questions. The event demystifies AI/ML and provide the fundamentals needed to be a leader in the era of big data.

We provide the data, you answer a pressing clinical question—all in two intense, exciting days.

2019 Event Details

The 2019 Health Tech Datathon took place on Friday, May 10, and Saturday, May 11, at NYU Langone’s Kimmel Pavilion.

For this event, we used the Medical Information Mart for Intensive Care (MIMIC) critical care database.

Agenda

Day 1: Friday, May 10

8:30–9:30AM: Check-In
9:30–9:45AM: Welcome Address
9:45–10:45AM: Introduction to the Datasets and Tools
10:45–11:00AM: Team Introductions
11:00AM–5:45PM: Datathon
5:45–6:00PM: Closing Remarks

Day 2: Saturday, May 11

8:30–9:15AM: Check-In
9:15–9:30AM: Welcome Address
9:30AM–3:00PM: Datathon
3:15–4:45PM: Team Presentations
4:45–5:15PM: Judging
5:15–5:30PM: Winner Announced
5:30–7:00PM: Cocktail Reception

Examples of Successful Clinical Questions and Publications

Examples of prior successful clinical questions include the following:

• predicting the next serum creatinine based on vital sign time series and treatments
• predicting the next serum lactate based on vital sign time series and treatments
• epidemiology of low-acuity patients who die in the intensive care unit (ICU)
• epidemiology of high-acuity patients who survive
• circadian variation in blood glucose
• optimizing medication dosing based on serum levels
• distribution of “normal” urine output and prediction of individualized goals for urine output based on serum creatinine and patient comorbidities
• outcomes in people over 80 admitted to the ICU
• does hemoglobin concentration affect the association between PaO2 and outcome in ventilated patients?
• does gender influence the outcome of sepsis?

Examples of publications generated from previous datathons using the MIMIC dataset include the following:

“Transthoracic echocardiography and mortality in sepsis: Analysis of the MIMIC-III database” in Intensive Care Medicine

“One-year mortality after recovery from critical illness: A retrospective cohort study” in PLoS One

“Potential adverse effects of broad-spectrum antimicrobial exposure in the intensive care unit” in Open Forum Infectious Diseases

“Comparing deep learning and concept extraction based methods for patient phenotyping from clinical narratives” in PLoS One

“Right ventricular function, peripheral edema, and acute kidney injury in critical illness” in Kidney International Reports

“A comparative analysis of sepsis identification methods in an electronic database” in Critical Care Medicine

“Effect of boarding on mortality in ICUs” in Critical Care Medicine

“Severity of illness scores may misclassify critically ill obese patients” in Critical Care Medicine

“The effect of ARDS on survival: Do patients die from ARDS or with ARDS?” in Journal of Intensive Care Medicine

“The association between sodium fluctuations and mortality in surgical patients requiring intensive care” in Journal of Critical Care

“Management of atrial fibrillation with rapid ventricular response in the intensive care unit: A secondary analysis of electronic health record data” in Shock

“Quantifying the mortality impact of do-not-resuscitate orders in the ICU” in Critical Care Medicine

“Understanding vasopressor intervention and weaning: Risk prediction in a public heterogeneous clinical time series database” in Journal of the American Medical Informatics Association

“Proton pump inhibitors are not associated with acute kidney injury in critical illness” in The Journal of Clinical Pharmacology

“The organizational structure of an intensive care unit influences treatment of hypotension among critically ill patients: A retrospective cohort study” in Journal of Critical Care

“Admission peripheral edema, central venous pressure, and survival in critically ill patients” in Annals of the American Thoracic Society

“Peripheral edema, central venous pressure, and risk of AKI in critical illness” in Clinical Journal of the American Society of Nephrology

Datathon Timeline

April 27, 2019

Our steering committee announced the eight clinical questions chosen for the datathon.

May 5, 2019

Our steering committee provided a tentative team formation structure. Teams will comprise clinicians and data scientists of varying skill levels and backgrounds. Teams can be modified until the event.

May 9, 2019

The eight selected teams will present their questions at the Health Tech Symposium.

Sessions at the symposium focused on data science and AI/ML. Stay tuned for additional information.

May 10–11, 2019

NYU Langone Health Tech Datathon

Datathon Logistics

At the Datathon, mentors from the Massachusetts Institute of Technology (MIT), along with NYU’s experts, will share their expertise in data science and the process.

Teams will query the MIMIC database to answer the eight selected clinical questions.

A winning team will be selected at the end of the event. However, the work conducted by all teams has the potential to be the foundation for research papers or quality improvement projects.

See a video summarizing a recent datathon held in France in collaboration with MIT.

Datathon Partners

We’re excited to partner with MIT and Google for the 2019 Health Tech Datathon.

MIT Critical Data is a collaborative group founded by the Laboratory of Computational Physiology (LCP) and the Computer Science and Artificial Intelligence Laboratory at the Massachusetts Institute of Technology (CSAIL). MIT Critical Data envisions a learning, data-driven health care system devoid of the barriers between practice and knowledge generation.

Google will be providing access to their BigQuery platform and Health Cloud Services.