Lung Cancer Survival Prediction via Machine Learning Regression, Classification, and Statistical Techniques
Problem
Predicting survival time in months for lung cancer patients is inaccurate for longer survival horizons, since prior models performed well only for survival under 6 months.
Why it matters
Reliable survival prognosis helps guide treatment decisions for lung cancer patients, a major cause of cancer death.
What the AI did
Regression models combined with a Random Forest classification model predicted survival time from de-identified SEER database records, using ANOVA-selected factors and evaluating via RMSE and confusion matrices.
Results
Random Forests performed best for survival times ≤6 and >24 months with RMSE 10.52 (abstract truncated thereafter)
Impact
The model improved prediction across survival horizons on retrospective SEER data; clinical validation or adoption is not stated.
Record
| Venue | IEEE International Symposium on Signal Processing and Information Technology (ISSPIT) |
|---|---|
| Published | 2018 |
| Work type | conference-paper |
| Authors | James A. Bartholomai, Hermann B. Frieboes |
| Citations | 61 (OpenAlex, at import time) |
| References | 31 |
| Model or method | Random Forests, Gradient Boosted Machines, general Linear Regression |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1109/isspit.2018.8642753 |
| Open access | green |
| Keywords | machine learning, lung cancer, survival prediction, SEER, random forests, regression |
Importance rating
significant — thousands to millions affected
Survival prognosis guides treatment for lung cancer patients; abstract shows improved prediction across horizons on retrospective SEER data but states no scale or clinical adoption.