| Title: | Risk-prediction models for developing and/or dying of lung cancer |
| Lead Institution: | University College London |
| Principal investigator: | Dr Thomas Callender |
Application 68073
WARNING: the interactive features of this website use CSS3, which your browser does not support. To use the full features of this website, please update your browser.
About
Lung cancer is the most common cause of death from cancer worldwide. Screening for lung cancer using a chest CT scan could reduce lung-cancer deaths by one-fifth. However, the current criteria to determine who should have screening leave out about half of those who go on to develop lung cancer. This includes younger smokers who may be at greater genetic risk of lung cancer. At the same time, those with a low risk of lung cancer have been shown to derive no benefit from screening, whilst being put at risk from the harms associated with unnecessary screening. These harms include surgery for patients who turn out not to have lung cancer. Genetic factors explain about 18% of the variation between people in the risk of developing lung cancer. Combining known genetic factors with other factors such as smoking history may allow us to more accurately select those who would benefit from screening. The first aim of this project is to test the performance of a gene-based risk model for lung cancer that was developed in a US study. We will then compare how well this model performs in determining who will and won't develop lung cancer against other risk calculators and criteria for lung cancer screening. The age at which lung cancer screening should start is unclear. Most risk calculators for lung cancer are only designed for use in those aged 55 and over. However, there are increasing calls to lower the age to 50. In other cancers, genetic-based risk calculators that predict the age at which an individual might develop cancer have been created. Meanwhile, few risk calculators have been developed or tested amongst those of a non-European ancestry or for never-smokers. Finally, machine learning or AI methods may be able to improve the accuracy of such risk calculators. Consequently, this project will tackle these areas - age-of-onset of screening, risk prediction models in different ethnicities, and the use of machine learning in lung cancer risk prediction models. The public health impact of this project will be to better understand the most effective methods of selecting those for lung cancer screening and to improve the equity of lung cancer screening programmes in diverse racial groups. The project duration is estimated to be 36 months.1 Publication
| Pub ID | Title | Author(s) | Year | Journal |
|---|---|---|---|---|
| 9429 | Assessing eligibility for lung cancer screening using parsimonious ensemble machine learning models: A development and validation study | Thomas Callender (+6) | 2023 | PLOS Medicine |
Enabling scientific discoveries that improve human health