TABLE 1.
LOO‐R 2 prediction performance results for all machine learning (ML) models when evaluating process parameters, and features from cytokine and nuclear magnetic resonance (NMR) media analysis at day 6 or day 4
| LOO‐R 2 | ML | ||||||
|---|---|---|---|---|---|---|---|
| Response/predictors | SR | RF | GBM | CIF | LASSO | PLSR | SVM |
| Ratio of CD4 to CD8 TN + TCM cells | |||||||
| PP + N4 | 99% | 86.8% | 96.3% | 84.5% | 88.6% | 92.5% | 88.5% |
| PP + N6 | 99% | 73.6% | 95.9% | 70.1% | 81.0% | 95.8% | 79.7% |
| PP + S6 | 99% | 87.1% | 99.9% | 83.4% | 87.2% | 97.9% | 86.8% |
| PP + S6 + N6 | 99% | 85.5% | 95.3% | 83.4% | 92.9% | 99.7% | 90.5% |
| Total live CD4+ TN + TCM cells | |||||||
| PP + N4 | 97% | 67.0% | 93.6% | 69.3% | 34.3% | 90.1% | 75.5% |
| PP + N6 | 96% | 45.9% | 92.6% | 51.2% | 42.8% | 92.1% | 79.4% |
| PP + S6 | 98% | 71.4% | 99.9% | 75.0% | 74.9% | 80.0% | 75.5% |
| PP + S6 + N6 | 98% | 68.2% | 95.6% | 74.4% | 72.5% | 81.7% | 77.0% |
| Total live CD8+ TN + TCM cells | |||||||
| PP + N4 | 93% | 4.7% | 44.4% | 9.2% | 1.2% | 65.1% | 9.1% |
| PP + N6 | 86% | 2.0% | 29.9% | 15.8% | 28.5% | 63.3% | 30.6% |
| PP + S6 | 93% | 7.8% | 28.0% | 15.1% | 76.2% | 98.4% | 49.8% |
| PP + S6 + N6 | 93% | 0.3% | 32.7% | 9.8% | 51.5% | 96.4% | 37.8% |
Notes: ML models' prediction performance is measured as the leave‐one‐out cross‐validated R 2 (LOO‐R 2) while SR prediction performance is measured as R 2 of the ensemble prediction where the ensemble is composed of diverse models with complexity constrained. Predictors evaluated: (PP) Process parameters, (N) NMR, (S) Cytokines measured at day 4 or 6. Maximum R 2 within each ML method are shown in bold.