Akihiko Nishimura
Johns Hopkins University, MD·Mathematics & Statistics
Hospitals, biobanks and patient registries now hold genetic and clinical records on huge numbers of people. Combined, these records could reveal which treatments work best for which patients.
- Fields
Mathematics & Statistics
Public Health & Epidemiology
Computer Science
- Topics
precision medicine
Bayesian hierarchical models
transfer learning
- Grants
- 1 known grant, through Aug 2030
What they work on
- Develop statistical methods that combine Bayesian hierarchical models, biomedical ontologies and high-dimensional transfer learning to pool genetic and clinical data from sites that differ in patients, data coding and practice.
- Scale these methods to modern biobank and health-record sizes by pairing fast algorithms with hardware optimization, including graphics processing units, and release them as open-source software.
- Apply the framework to genetic risk prediction for under-represented populations, using data combined from multiple sources.
- Estimate how the effects of type 2 diabetes treatments differ across patient subgroups in comparative effectiveness studies, using a federated network of healthcare databases.
More about the research
Combining them is hard, though. Each site serves different patient populations, codes its data differently and follows different clinical practices. The data are also so large that standard statistical tools cannot keep up. This project builds statistical and computational methods to pool information across many such sources.
Why it matters
Patients vary widely in how they respond to treatment, so choosing the right therapy for each person requires evidence pooled from many data sources. These methods could improve genetic risk prediction for under-represented populations. They could also help identify which type 2 diabetes treatments work best for which patients. The open-source software would be available to researchers and clinicians working with large health datasets.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Tianxi Cai
Harvard University, MA·Mathematics & Statistics
Develop semi-supervised risk prediction methods that combine noisy electronic health record surrogates, such as billing codes, with a small set of chart-reviewed outcome labels, even when predictors are numerous.
Mei Liu
University of Florida, FL·Medicine & Clinical Research
Develop methods to build the best individualized cohort of similar past patients from EHR data, as the training set for a personalized AKI risk model for each new hospital patient.
Catherine Sun
University of California-San Francisco, CA·Medicine & Clinical Research·Early-career
Develop an automated method to identify patients with proliferative diabetic retinopathy in electronic health records, combining structured data such as diagnosis codes, medications and labs with unstructured clinical notes.
The school
Johns Hopkins University: 100 labs on PI Pages, $4.1B a year on research, 622 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.