Fig. 1
From: Predicting accrual success for better clinical trial resource allocation
The number (a) and percentages (b) of the three type of trials over time. To access the generalizability of our models over time, we used a prospective cross-validation design (c) where historical data were used as discovery data (yellow) for model training and selection, whereas future data were used as validation data (red). x-axis represents time.
