Li-Wei H Lehman
Massachusetts Institute of Technology, MA·Medicine & Clinical Research
When a patient's lungs fail suddenly and severely — a condition called acute respiratory distress syndrome, or ARDS — doctors must decide, hour by hour, whether to give steroids, whether to paralyze the breathing muscles with drugs so a ventilator can take over, and how hard to push air into the lungs. Many of these choices rest on thin or conflicting evidence, a gap that became painfully clear during COVID-19, when ARDS was a leading cause of death.
- Fields
Medicine & Clinical Research
Computer Science
Public Health & Epidemiology
- Topics
acute respiratory distress syndrome
critical care
intensive care unit
- Grants
- 2 known grants, through Aug 2027
What they work on
- Estimate what would happen under different treatment strategies for neuromuscular blockade and steroids in ARDS by emulating target trials in intensive care records, using g-methods that correct for the fact that sicker patients receive more aggressive care.
- Compare patient outcomes under alternative thresholds for driving pressure and mechanical power — measures of how forcefully a ventilator inflates the lung — across the MIMIC-IV, eICU and AmsterdamUMCdb databases to test whether findings hold across countries and centers.
- Develop machine learning methods that turn routinely collected, continuously changing ICU measurements into dynamic ARDS phenotypes for predicting prognosis and for building personalized dynamic treatment regimes.
- Test whether early, sustained steroid use improves outcomes compared with delayed, abbreviated or no-steroid regimes across a range of doses in ARDS patients in MIMIC-IV and eICU.
- Evaluate rules that trigger steroid initiation based on evolving markers of disease severity, and estimate how steroid strategies differ for patients who have ARDS with sepsis versus ARDS alone.
More about the research
Randomized trials cannot test every possible sequence of decisions. This research asks whether large archives of intensive care data, analyzed with careful causal-inference methods, can answer those questions instead, and whether machine learning can tell apart ARDS subtypes that respond differently to treatment.
Why it matters
ARDS kills a large share of the patients who develop it, and clinicians routinely use treatments whose benefit or harm is genuinely unknown. Rigorous causal analyses of existing ICU records could show which patients benefit from steroids or muscle paralysis and at what ventilator settings, guide the design of future randomized trials, and give the critical care field shared code and cohort templates that raise the quality of evidence drawn from observational data.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Rishikesan Kamaleswaran
Duke University, NC·Biomedical Engineering
Discover 'physiomarkers' in continuous ICU monitor data, including high-fidelity waveforms, that predict the onset of sepsis in patients who develop secondary infections after ICU admission.
Jacqueline Stocking
University of California-Davis, CA·Medicine & Clinical Research·Early-career
Optimize and validate an automated, EHR-based model predicting postoperative respiratory failure, using surgical patient data from 2014 to 2021 and pre- and intra-operative risk factors.
Romain Pirracchio
University of California-San Francisco, CA·Medicine & Clinical Research
Train machine learning models on electronic health record data from three health systems to estimate each patient's individual treatment effect from antibiotics, fluid resuscitation and vasopressors in suspected community-onset lung sepsis.
The school
Massachusetts Institute of Technology: 25 labs on PI Pages, $1.1B a year on research, 692 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.