Digital antibiotic susceptibility testing: an extra weapon in the fight against antibiotic resistance

The growing spread of antibiotic resistance is one of the most serious threats to public health worldwide. Bacterial strains that are resistant to most antibiotics are emerging with increasing frequency – more so in Italia than elsewhere – giving Italia a European ‘record’ that is certainly not one to be proud of. The causes are linked to the excessive and inappropriate use of antibiotics; the overuse of broad-spectrum antibiotics, which should instead be replaced more often by more targeted ones; and the selective pressure caused by the excessive use of antibiotics in livestock farming.

Compounding the situation is the time factor, which is particularly critical in the case of the most serious infections. Minimising the time taken to start the most effective treatment can significantly improve the course of the illness, particularly in elderly or frail patients or those in critical condition, thereby making a decisive contribution to improving clinical outcomes, shortening the duration of hospital stays whilst also reducing their costs, lowering mortality associated with the most serious infections whilst limiting the use of broad-spectrum antibiotics, and finally combating the emergence of antibiotic resistance.

The severity associated with the most serious infections is a consequence of what is known as the ‘golden hour’. Several studies estimate a 5–10 per cent increase in mortality for every hour’s delay in initiating effective treatment in patients suffering from septic shock. Consequently, any delay in identifying the most effective antibiotic treatment can worsen clinical outcomes. Therefore, predicting the response to antibiotics early, in a personalised and effective manner, can significantly alter the clinical course of many infections.

A recent Italian study shows that machine learning can be a valuable aid in the timely and personalised selection of the most effective antibiotics.

Using each patient’s individual clinical and microbiological data, the researchers have selected a predictive algorithm known as the ‘digital antibiogram’, which, with an accuracy of nearly 90 per cent, predicts sensitivity to different antibiotics on a patient-by-patient basis and can support clinical decision-making, although the final decision remains with the doctor. The results are processed at least 48 hours in advance of the time required to obtain a standard antibiotic susceptibility test.