Improving the diagnostic performance of MUNIN by remodelling of the diseases

Abstract

The paper describes a revision of the structure of MUNIN, a causal probabilistic network that specifies a stochastic model of the relations between a range of neuromuscular diseases and the findings associated with these diseases. The stochastic model was revised 1) to achieve a more flexible specification of the anatomical distribution of findings associated with the diseases, 2) to allow diagnosis of diseases or groups of diseases (e.g. polyneuropathies and motor neuron diseases) previously lumped under the common concept of diffuse neuropathy and 3) to model the correlation between carpal tunnel syndrome on the left and right side. Minor adjustments were also made to some of the conditional probabilities used to specify the pathophysiology of the diseases. The diagnostic capability of the revised model was evaluated by letting MUNIN diagnose 30 cases. The evaluation showed that the revised model performed better with a sensitivity of 94% and a specificity also of 94%.