Use of artificial intelligence for the automatic assessment of left ventricular ejection fraction by oncology staff in chemotherapy patients

Abstract Funding Acknowledgements Type of funding sources: None. Background The calculation LV ejection fraction (LVEF) by transthoracic echocardiography is pivotal in detecting cancer therapy–related cardiac dysfunction. Referrals for LVEF estimation pre- and post-chemotherapy occupy significant amount resources laboratories increase service deliverance. Novel handheld ultrasound devices (HUDs) can provide echocardiographic images at the point care with diagnostic image quality. Recently, artificial intelligence (AI) technology enabled development algorithms real-time guidance probe to acquire optimal heart calculate automatically. Purpose To evaluate feasibility accuracy oncology staff using an AI HUD. Methods We studied 115 patients referred estimation. All were scanned a cardiologist standard (SE) systems biplane Simpson’s rule was used as reference standard. A brief training on basics use HUD provided before study. Then, each patient independently cardiologist, senior oncologist, resident, nurse AUTO-GUIDANCE AUTO-GRADING applications (Figure 1) apical 4-chamber 2-chamber views heart. automatically calculated device autoEF algorithm. Method agreement assessed Pearson’s correlation Bland-Altman analysis. detection impaired LVEF<50%, commonly cut-off deferring chemotherapy, calculated. Results Diagnostic acquisition possible 96% cases 94% 93% junior oncologist 89% nurse. between SE-EF 2A) excellent (r=0.90), good (r=0.79), (r=0.82) (r=0.84), p<0.001 all. plots 2B) showed small underestimation algorithm compared all operators. There bias −2.1% −3.5% −2.2% −2.3% (p<0.001 all). Detection feasible sensitivity 95% specificity cardiologist; 86% oncologist; 91% 87% Conclusions Calculation automated selected population. accuracy. These findings show clinical potential improve quality patients.