Caleb Miles
Columbia University, NY·Mathematics & Statistics
Clinical trials for psychiatric and substance use disorders are hard, slow and expensive to run, so they are often too small. Small trials can miss modest treatment effects that still matter for public health.
- Fields
Mathematics & Statistics
Public Health & Epidemiology
Medicine & Clinical Research
- Topics
biostatistics
causal inference
data fusion
- Grants
- 1 known grant, through May 2029
What they work on
- Develop simple parametric and semiparametric efficient estimators of the average treatment effect in multi-study, multi-outcome data fusion, using related follow-up measurements as proxies for outcomes missing in some trials.
- Develop estimators of conditional average treatment effects and optimal individualized treatment rules that match patients to treatments based on demographic and clinical characteristics in fused multi-outcome data.
- Build sensitivity analyses for average and conditional treatment effect estimates when the untestable transport assumption, that effects carry over between studies, does not hold.
- Apply these estimators to harmonized trials of medications for opioid use disorder (MOUD) with multiple outcomes to increase power to detect treatment effects.
- Create and share user-friendly software that implements all the data fusion, estimation and sensitivity analysis methods.
More about the research
They also lack the data needed to learn which treatment works best for which patient, the core of personalized medicine. One fix is data fusion: pooling individual patient data from several similar studies. But trials often measure different outcomes at follow-up. Studies that lack the exact outcome of interest are usually left out, which wastes useful information. This project develops statistical methods to combine trials even when they measure different but related outcomes, applied to treatment for opioid use disorder.
Why it matters
People with opioid use disorder could benefit if existing trial data were used more fully. Methods that combine trials with different outcomes could detect modest but important medication effects and help match medications for opioid use disorder to individual patients rather than using one-size-fits-all treatment. The free software could let other researchers apply these methods to psychiatric and substance use disorder trials.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Alyssa Carlson
University of Missouri-Columbia, MO·Social Science·Early-career
Establish identification in triangular random coefficient models with continuous endogenous variables using a general control function that does not require scalar heterogeneity, the assumption of a single hidden first-stage factor.
Barbara Andraka-Christou
University of Central Florida, FL·Public Health & Epidemiology
Build a public, year-by-year database of state statutes, regulations and executive orders (2014–2024) meant to promote medications for opioid use disorder after ED overdose visits, using expert-guided searches of NexisUni legal software.
Ashish Thakrar
University of Pennsylvania, PA·Medicine & Clinical Research·Early-career
Conduct a dose-finding study to identify the maximum tolerated methadone loading dose for people with opioid use disorder who use fentanyl, to guide future studies.
The school
Columbia University: 89 labs on PI Pages, $1.4B a year on research, 625 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.