Schematic representation of parameters used in machine learning for diagnosis of AA. (A) Representative bone marrow biopsy (trephine), low power (×4, ×10, ×20 lens objectives), showing hypocellular marrow with absent/reduced hematopoietic cells. (B) Example of a TL report showing a representative patient with a TL in <first centile. (C) The top 7 parameters used in machine learning and the parameters that showed statistical significance in logistical regression analysis. Refer to Figure 4 and supplemental Table 4 in Gutierrez-Rodrigues et al.1 (D) Spectrum of inherited and acquired BMFS, with age distribution and overlap with hypoplastic MDSs. (E) Additional parameters for future model building and potential future utility of AI in AA. HbF, fetal hemoglobin. Professional illustration by Patrick Lane, ScEYEnce Studios.
