Tejal Somvanshi
Researchers have developed a machine-learning model called PreCSenM to measure cellular senescence in lung adenocarcinoma. The model is designed to turn gene-expression data into a consistent senescence score.
Cellular senescence is the state in which damaged cells stop dividing. In cancer biology, it is difficult to measure consistently because different datasets, cell types and experimental conditions can produce uneven signals.
The research team assembled 888 transcriptomic profiles across different cell types and senescence-inducing conditions, including replicative exhaustion, oncogene activation and drug treatment. They used the Boruta feature-selection algorithm to identify predictive genes and build a cellular-senescence gene signature.
The team benchmarked ten machine-learning algorithms, including logistic regression, support vector machines, random forest, XGBoost, partial least-squares regression and artificial neural networks. PreCSenM performed better than existing senescence-quantification approaches on validation metrics reported in the source article.
Karmactive has previously covered cancer research and treatment and drug-approval stories in oncology such as pancreatic cancer treatment coverage. This story is about the measurement layer: before a score can be used widely, researchers need a method that can compare samples without changing the meaning of the result.
