Fig 1.
Machine learning pipeline for the prediction of Alzheimer’s disease conversion.
Fig 2.
Censored and uncensored data distribution per year in early and late MCI individuals.
Table 1.
Number of missing values in features of eMCI and lMCI datasets.
Fig 3.
Distribution of censored and uncensored data in eMCI and lMCI groups.
Fig 4.
Balancing the training set of eMCI group.
Fig 5.
Comparison of survival curves of eMCI and lMCI groups showing varying survival probabilities.
Table 2.
Data statistics of eMCI and lMCI groups in this study.
Table 3.
Best performing hyperparameters obtained using Grid Search-CV.
Fig 6.
Heatmap showing performance of models measured by C-Index and IBS for early MCI.
Fig 7.
Heatmap showing performance of models measured by C-Index and IBS for late MCI.
Fig 8.
Feature importance results of eMCI dataset.
The error bars show the standard deviation, and each bar indicates the mean score.
Fig 9.
Feature importance results of lMCI dataset.
The error bars show the standard deviation, and each bar indicates the mean score.
Fig 10.
Predicted survival estimates for subjects with (a) progressive eMCI and (b) non-progressive eMCI. The red line refers to the actual event times for progressive/uncensored patients and the actual censoring time for non-progressive/censored patients.
Fig 11.
Predicted survival estimates for subjects with (c) progressive lMCI and (d) non-progressive lMCI. The red line refers to the actual event times for progressive/uncensored patients and the actual censoring time for non-progressive/censored patients.
Table 4.
RSF’s performance before and after applying data balancing.