A causal probabilistic network for optimal treatment of bacterial infections

Abstract

AbstractÐThe fatality rate associated with severe bacterial infections is about 30 percent and appropriate antibiotic treatment reduces it by half. Unfortunately, about a third of antibiotic treatments prescribed by physicians are inappropriate. We have built a causal probabilistic network (CPN) for treatment of severe bacterial infections. The net is based on modules, each module representing a site of infection. The general configuration of a module is as follows: Major distribution factors define groups of patients, each of them with a definite prevalence of infection caused by a given pathogen. Minor distribution factors multiply the likelihood of one pathogen, without changing much of the prevalence of infection. Infection caused by a pathogen causes local and generalized signs and symptoms. Antibiotic treatment is appropriate if it matches the susceptibility of the pathogens in vitro and appropriate treatment is associated with a gain in life expectancy. This is balanced against the cost of the drug, side effects, and ecological damage, to reach the most cost effective treatment. The net was constructed in such a way that the data for the conditional probability tables will be available, even if it meant sometimes giving up on fine modeling details. For data, we used large databases collected by us in the last 10 years and data from the literature. The CPN was a convenient way to combine data from databases collected at different locations and times with published information. Although the net is based on detailed and large databases, its calibration to new sites requires data that is available in most modern hospitals.

A Probabilistic Network for Fusion of Data and Knowledge in Clinical Microbiology

“…These assumptions are hardly controversial, but they dictate the basic topology of the CPN. How this translation from the assumption to the topology is performed has been described in previous publications [1,2,11]. The starting point for the calculation of coverage is the distribution of probabilities over the different pathogens that may cause the infection at each site.…”

“…It is also very interesting that although the four intermediate factors are statistical constructs, they probably have a high degree of physical reality. For example, the fever factor "Fact fever" can essentially be mapped onto the plasma level of interleukin-6 [11], one of the factors that are known to mediate the immune system's inflammatory systemic response.…”

“…This can be avoided by introducing an intermediate layer of variables. The four intermediate factors were found by factor analysis, and, together, they explain more than 80% of the variance in the dependent variables [11]. The data required for this exercise were obtained from the Rabin bacteraemia database and normal values from the literature.…”

“…This long-standing interest in antibiotic therapy is not without reason: infections are a frequent cause of death, but the risk of death can be substantially reduced by appropriate antibiotic therapy [13]. In current clinical practice, about a third of patients with moderate to severe infections receive inappropriate antibiotic therapy resulting in increased morbidity, extended hospital stay, and increased mortality [11]. Of more importance for the role of decisionsupport systems, it can be noted that testing of decision-support systems indicates that antibiotic therapy can be improved beyond current clinical practice.…”

An Anthology of Probabilistic Models for Medical Informatics

“…These networks can provide insight into the way various factors and mechanisms interact to influence the outcome of a disease in a patient. The applicability of BNs to medicine has been investigated in a number of areas, including bone-marrow transplantation [150], intensive care [9], electromyography [2], clinical microbiology [90], and oncology [46].…”