Nat Commun: Using game theory to reveal the relationship between clinical information differentiation and antibiotic abuse

Resistance to antibiotics has significant public health implications worldwide.
antibiotic-resistant infections increase mortality, prolong hospital stays and increase treatment costs.
when a patient is suspected of a bacterial infection, biological specimens are usually collected, cultured, and tested for antibiotic sensitivity.
, the results are known within a few days.
results, the doctor can prescribe the right treatment.
, these days may be critical for treatment, so doctors will give initial, experienced antibiotic treatment before identifying pathogenic bacteria.
options for empirical treatment can be based on the doctor's informal assessment of the likelihood of different types of infection, on the patient's medical symptoms, clinical environment and the results of immediate diagnostic tests, or on the output of the decision support system.
the problem of drug resistance caused by antibiotic abuse is a typical public welfare problem, which is suitable for game theory analysis.
, the researchers designed a model in which doctors weigh whether antibiotics should be prescribed, taking into account their future effectiveness.
doctor's decision depends on the probability of bacterial infection before clear laboratory results come out.
that doctors' balanced decision-making rules on antibiotic prescriptions are not socially optimal.
further demonstrated that discrete processing of information provided to doctors could ease the gap between their balanced decision-making and the social optimality of antibiotic prescriptions.
complexity of the problem, the effectiveness of discreteness depends entirely on the type of information that doctors can use to determine the nature of the infection.
this has been demonstrated in theoretical distribution and clinical data sets.
, the results provide a game-based guide to the optimal output of current and future antibiotic prescription decision support systems.