A machine learning algorithm can predict lung cancer risk and determine who requires lung cancer screening utilizing data on people's age, length of smoking, and amount of cigarettes smoked daily.
The most frequent type of cancer that results in mortality globally is lung cancer, which has a terrible prognosis without early identification.
However, the number of lung cancer fatalities might be cut in half by screening people who are most at risk, but it is unclear how to identify this group best.
Only a handful of the 17 characteristics the current standard-of-care model needs for predicting lung cancer risk are frequently found in electronic health data.
How did they run the study?
Thomas Callender of University College London and colleagues calculated the likelihood that a person would get lung cancer and pass away over five years, using a machine learning model that considered three indicators, including age, smoking history, and pack-years.
The researchers tested the new model on a third data set from the Prostate, Lung, Colorectal, and Ovarian Screening Trial (PLCO).
The model accurately predicted lung cancer fatalities with a sensitivity of 85.5% and lung cancer incidence with a sensitivity of 83.9%. Each model variant exhibited a higher sensitivity at equivalent specificities than the present risk prediction algorithms.
Overall, the team used machine learning to simplify significantly the process of determining who is at high risk and presented a method that may mark an exciting first step toward the broad use of individualized screening to find illnesses early.
In his conclusion, Callender states that screening people at a high risk of acquiring lung cancer can prevent fatalities.