randomized trials
Experiments that randomly assign subjects to treatment and control groups to establish causal effects; the general scientific form of A/B tests.
As defined in Pedro Domingos on Machine Learning and the Master Algorithm
Where it comes up · 2
- Jonathan Skinner on Health Care Costs, Technology, and Rising Mortality3 claims
- Pedro Domingos on Machine Learning and the Master Algorithm2 claims
Most important claims · 5 of 5Search all claims for “randomized trials” →
- It is a common misconception that machine learning can only discover correlation; ML can and does discover causation by performing experiments such as A/B tests and randomized trials, and can go further than asking whether a drug works to identify exactly who it works for by examining all the other variables involved.Pedro Domingos on Machine Learning and the Master Algorithm
- Randomized trials are often better than observational data but still contain unobservables, and many randomized trial results fail to generalize precisely because we don't understand the underlying causal mechanism; observational similarity for the data you happen to have does not establish that subjects are truly similar at, say, the cellular level.Pedro Domingos on Machine Learning and the Master Algorithm
- Individual testimonials about treatments saving a life or therapy helping are unreliable because there is no counterfactual—Roberts's shoulder improved after he stopped physical therapy, illustrating that without knowing what would have happened otherwise, one cannot attribute outcomes to a treatment, which is why randomized trials are valuable.Jonathan Skinner on Health Care Costs, Technology, and Rising Mortality
- A randomized trial can show huge benefit for a carefully selected group of complication-free patients treated by the best doctors at the best centers; physicians read this in journals like NEJM or JAMA, conclude the treatment works, and the technology diffuses unevenly and widely—yet the actual benefit for the marginal patient outside those ideal conditions may be far smaller.Jonathan Skinner on Health Care Costs, Technology, and Rising Mortality
- Despite current implementation difficulties, electronic health records have tremendous potential because a different doctor can access a patient's full records, physicians' adherence to needed tests can be monitored, and computers can detect patterns across the roughly 30,000 randomized trials done yearly to prompt physicians toward rare diagnoses no human could remember.Jonathan Skinner on Health Care Costs, Technology, and Rising Mortality