Recent developments in probabilistic reasoning in expert systems

In collaboration with clinicians, various aids to decision-making based on statistical analysis of databases have been developed and evaluated. The paper concerns the development and implementation of rigorous probabilistic means of handling uncertainty in expert systems which use a causal network representation of knowledge. Future work is concerned with uniting these two strands to enable expert opinion and patient databases to be smoothly integrated to form knowledge-bases which may be interrogated and improved in an efficient but statistically appropriate way