A deep learning model successfully predicted the lung cancer survival period with an accuracy of 71.18%, outperforming previous machine learning models, according to the results of a study published in the International Journal of Medical Informatics.

“Early detection and prediction of depth of survivability from cancer can help both patients and healthcare professionals better manage costs, treatment intensity and time spent around medical care,” the authors wrote. The aim of this study was to characterize a deep learning approach to predict the survival period of patients with lung cancer.

The study used data from the Surveillance, Epidemiology, and End Results (SEER) program. The deep learning models included data preprocessing using categorical and quantitative variables. The researchers segmented survival time into ≤6 months, 0.5 to 2 years, and >2 years. They used 3 different deep learning architectures to model lung cancer survival: artificial neural networks (ANN), recurrent neural networks (RNN), and convolutional neural networks (CNN). The investigators compared the results with previously used machine learning models, including the classification models and regression models.


Continue Reading

The deep learning models outperformed the traditional machine learning models. The best performance was with CNN, which achieved a root mean square error (RMSE) of 13.5% and an R2 value of 50.66%. This was closely followed by the RNN model, at 13.58% RMSE and an R2 of 50.09%, and the ANN model with an RMSE of 13.89% and an R2 of 47.78%. The traditional machine learning models achieved RMSEs that ranged from 14.87% to 15.5% and an R2 of 35.04% to 40.19%.

Overall, the classification approach using deep learning models achieved an accuracy of 71.18%.

“The performance of survival models could be improved through the addition of temporal data to enhance the models built that help in sustaining predictions over longer periods and increase the prediction accuracy for cases with long survival period,” the authors wrote.

The authors concluded that “the improved outcomes of this study help in developing an early planning mechanism that provides a base to plan treatments, finances and resources required for a patient at the point of early diagnosis of a cancer patient.”

Reference

Doppalapudi S, Qiu RG, Badr Y. Lung cancer survival period prediction and understanding: Deep learning approaches. Int J Med Informatics. 2021;148:104371. doi:10.1016/j.ijmedinf.2020.104371