Hemopneumothorax detection through the process of artificial evolution - a feasibility study

Abstract Background Tension pneumothorax is one of the leading causes preventable death on battlefield. Current prehospital diagnosis relies a subjective clinical impression complemented by manual thoracic and respiratory examination. These techniques are not fully applicable in field conditions battlefield, where situational environmental factors may impair capabilities. We aimed to assemble device able sample, analyze, classify unique acoustic signatures hemothorax. Methods Acoustic data was obtained with simultaneous use two sensitive digital stethoscopes from chest wall an ex-vivo porcine model. Twelve second samples were in-house assembled stethoscope system during mechanical ventilation. The cavity injected increasing volumes 200, 400, 600, 800, 1000 ml air or saline simulate hemothorax, respectively. analyzed using multi-objective genetic algorithm that used develop optimal mathematical detector through process artificial evolution, cutting-edge approach intelligence discipline. Results dual developed achieved accuracy, sensitivity specificity ranging 64 100%, 63 respectively, classifying signal as associated hemothorax at fluid injection levels 400 more, regardless background noise. Conclusions present novel, objective for rapid potentially lethal injuries. With further optimization, such could provide real-time detection monitoring battlefield conditions.