Leading at the intersection of data science, clinical research, and technology
Tomasz Adamusiak is a physician-scientist with broad data science experience across pharma, government, and payer health services, focused on large language models, clinical AI, and healthcare outcomes. He holds an MD and a PhD in clinical immunology from the Medical University of Łódź, and was an Erasmus Fellow at the University of Erlangen-Nuremberg.
A ceiling-mounted radar, with no sensor worn or attached, recovers two waveforms: the chest wall moving with each breath, and the great vessels pulsing with each beat. Heart rate, respiratory rate and variability stay inside their reference ranges the entire time. The model tracks the phase relationship between the two waveforms rather than their absolute levels.
The waveforms are synthesised rather than recorded. The respiratory asymmetry, the sinus arrhythmia, the arterial pulse morphology and the phase-locked amplitude modulation are modelled from physiology; the ectopic carries a true compensatory pause. None of this uses patient data.
A protocol amendment goes out. Three agents work out whether this person qualifies and how the visits would actually fit their life — the patient’s agent, the study site’s, and the sponsor’s. The record never leaves the patient’s side: eligibility is evaluated where the data is already stored, and only the answer travels.
Illustrative exchange. Eligibility evaluated locally against the protocol rather than by shipping the record to a sponsor is the part that is not yet routine.
The Pooled Cohort Equations have set US statin thresholds since 2013, and they ask your race. The AHA’s PREVENT equations, published in 2024, drop the race term, add kidney function, and were fitted on 3.3 million people. Both are computed here from their published coefficients for the same patient. Each line is one simulated ten-year course; the dashed line is the 7.5% threshold that decides who is offered a statin.
Goff et al., 2013 ACC/AHA Guideline on the Assessment of Cardiovascular Risk. Circulation 2014;129:S49–S73. Khan et al., Development and Validation of the American Heart Association’s PREVENT Equations. Circulation 2024;149:430–449 — implementation validated against the AHA PREVENT calculator. Applying PREVENT would move 53% of US adults into a lower risk category (Diao et al., JAMA 2024;332:989–1000). PCE validated for ages 40–79, PREVENT 30–79, both without prior CVD.
“The best way to predict the future is to create it.”