Blood: Machine learning reveals the correlation between MDS genotypes and morphological characteristics.

Bone marrow growth abnormal syndrome (MDS) is a group of tumor diseases originating from hematopoietic stem cells, the main characteristics are abnormal bone marrow hematopoietic function, abnormal development of blood cells, performance of incurable blood cell reduction, hematopoietic failure, there is a high risk of conversion to acute leukemia, once known as pre-leukemia.
Has long been the gold standard for diagnosing MDS for morphological examination of blood and bone marrow cells, but this approach has limitations, such as uneven reliability of pathological assessments and lack of integration with genetic data.
events affect morphological characteristics, but the complexity of morphological and genetic changes makes it challenging to identify their associations.
study used machine learning techniques to identify symbic patterns between morphological features and genomic events to analyze new clinical subsequencies of MDS.
study sequenced 1,079 MDS patients and analyzed bone marrow morphological changes and other clinical features.
a total of 1929 individual cell mutations were identified.
prognostics of five populations with different morphological characteristics define five different morphological characteristics with unique clinical characteristics (Profile-1, Profile-2, Profile-3, Profile-4, and Profele-5).
77% of high-risk patients are concentrated in Profele-1.
all low-risk patients are concentrated in the remaining four types: Profile-2 shows a decrease in whole blood cells, Profile-3 shows an increase in monocytes, Profile-4 shows an increase in cytocytes, and Profile-5 manifests it as an abnormal growth of the red line.
these characteristics can also distinguish between patients with different prognosis.
patients with low-risk MDS were divided into 8 genetic characteristics associated with specific morphological characteristics (e.g., Signal-A has TAT2 mutations, Signature-B has TAT2 and SRSF2 mutations, and Signature-G has SF3B1 mutations).
the number of genes validated in a separate queue in a separate analysis, the correlation between six morphological/genetic characteristics was verified.
, the study shows that non-random, and even pathogenic, relationships between morphology and genotype can be identified to define clinical characteristics.
this is the first time that machine learning algorithms have been fully implemented to illustrate the potential intrinsic interdependence between genetic damage, morphology, and clinical prognostics in MDS properties.
.