A Bayesian Approach for Medical Inquiry and Disease Inference in Automated Differential Diagnosis
Citations Over Time
Abstract
We propose a Bayesian approach for both medical inquiry and disease inference, the two major phases in differential diagnosis. Unlike previous work that simulates data from given probabilities and uses ML algorithms on them, we directly use the Quick Medical Reference (QMR) belief network, and apply Bayesian inference in the inference phase and Bayesian experimental design in the inquiry phase. Moreover, we improve the inquiry phase by extending the Bayesian experimental design framework from one-step search to multi-step search. Our approach has some practical advantages as it is interpretable, free of costly training, and able to adapt to new changes without any additional effort. Our experiments show that our approach achieves new state-of-the-art results on two simulated datasets, SymCAT and HPO, and competitive results on two diagnosis dialogue datasets, Muzhi and Dxy.
Related Papers
- → The Bayesian New Statistics: Hypothesis testing, estimation, meta-analysis, and power analysis from a Bayesian perspective(2017)1,295 cited
- → Bayeswatch: an overview of Bayesian statistics(2002)30 cited
- → Comparing Bayesian Statistics and Frequentist Statistics in Serious Games Research(2021)8 cited
- → Reporting Bayesian Results(2020)7 cited
- → The Bayesian New Statistics (Public)(2017)1 cited