Using machine learning to guide targeted and locally-tailored empiric antibiotic prescribing in a children’s hospital in Cambodia

AbstractBackgroundEarly and appropriate empiric antibiotic treatment of patients suspected of having sepsis is associated with reduced mortality. The increasing prevalence of antimicrobial resistance risks eroding the benefits of such empiric therapy. This problem is particularly severe for children in developing country settings. We hypothesized that by applying machine learning approaches to readily collected patient data, it would be possible to obtain actionable and patient-specific predictions for antibiotic-susceptibility. If sufficient discriminatory power can be achieved, such predictions could lead to substantial improvements in the chances of choosing an appropriate antibiotic for empiric therapy, while minimizing the risk of increased selection for resistance due to use of antibiotics usually held in reserve.Methods and FindingsWe analyzed blood culture data collected from a 100-bed children’s hospital in North-West Cambodia between February 2013 and January 2016. Clinical, demographic and living condition information for each child was captured with 35 independent variables. Using these variables, we used a suite of machine learning algorithms to predict Gram stains and whether bacterial pathogens could be treated with standard empiric antibiotic therapies: i) ampicillin and gentamicin; ii) ceftriaxone; iii) at least one of the above.243 cases of bloodstream infection were available for analysis. We used 195 (80%) to train the algorithms, and 48 (20%) for evaluation. We found that the random forest method had the best predictive performance overall as assessed by the area under the receiver operating characteristic curve (AUC), though support vector machine with radial kernel had similar performance for predicting Gram stain and ceftriaxone susceptibility. Predictive performance of logistic regression, simple and boosted decision trees and k-nearest neighbors were poor in comparison. The random forest method gave an AUC of 0.91 (95%CI 0.81-1.00) for predicting susceptibility to ceftriaxone, 0.75 (0.60-0.90) for susceptibility to ampicillin and gentamicin, 0.76 (0.59-0.93) for susceptibility to neither, and 0.69 (0.53-0.85) for Gram stain result. The most important variables for predicting susceptibility were time from admission to blood culture, patient age, hospital versus community-acquired infection, and age-adjusted weight score.ConclusionsApplying machine learning algorithms to patient data that are readily available even in resource-limited hospital settings can provide highly informative predictions on susceptibilities of pathogens to guide appropriate empiric antibiotic therapy. Used as a decision support tool, such approaches have the potential to lead to better targeting of empiric therapy, improve patient outcomes and reduce the burden of antimicrobial resistance.Author summaryWhy was this study done?Early and appropriate antibiotic treatment of patients with life-threatening bacterial infections is thought to reduce the risk of mortality.In hospitals that have a microbiology laboratory, it takes 3-4 days to get results which indicate which antibiotics are likely to be effective; before this information is available antibiotics have to be prescribed empirically i.e. without knowledge of the causative organism.Increasing resistance to antibiotics amongst bacteria makes finding an appropriate antibiotic to use empirically difficult; this problem is particularly severe for children in developing country settings.If we could predict which antibiotics were likely to be effective at the time of starting antibiotic therapy, we might be able to improve patient outcomes and reduce resistance.What Did the Researchers Do and Find?We evaluated the ability of a number of different algorithms (i.e. sets of step-by-step instructions) to predict susceptibility to commonly-used antibiotics using routinely available patient data from a children’s hospital in Cambodia.We found that an algorithm called random forests enabled surprisingly accurate predictions, particularly for predicting whether the infection was likely to be treatable with ceftriaxone, the most commonly used empiric antibiotic at the study hospital.Using this approach it would be possible to correctly predict when a different antibiotic would be needed for empiric treatment over 80% of the time, while recommending a different antibiotic when ceftriaxone would suffice less than 20% of the time.What Do These Findings Mean?Using readily available patient information, sophisticated algorithms can enable good predictions of whether antibiotics are likely to be effective several days before laboratory tests are available.Algorithms would need to be trained with local hospital data, but our study shows that even with relatively limited data from a small hospital, good predictions can be obtained.Used as part of a decision support system such algorithms could help choose appropriate antibiotics for empiric therapy; this would be expected to translate into better patient outcomes and may help to reduce resistance.Such as a decision support system would have very low costs and be easy to implement in low- and middle-income countries.