Table 2.
Early warning methods.
| Type of analysis | Early warning parameter | Early warning level / cutoff level | Abbreviation | Data type |
|---|---|---|---|---|
|
Real-time analysis |
Data sequence numerical increase or decrease | Keep increasing for 3 consecutive data points | OBMN1N2⁎ | N1 N2 gene concentrations |
| Positive percentage change | > 40 % | PPCN1N2 | N1 N2 gene concentrations | |
| Positive percentage change of slope | > 200 % | PPCS200⁎ | N1 N2 gene concentrations | |
|
Post-factum analysis |
Mean + 2 standard deviation | Intersection values higher than mean + 2SD threshold | MSD (B1) ⁎ | N1 N2 gene concentrations and w/c ratio |
| Variance / mean | Intersection values higher than variance/mean threshold | VAM (B2) | ||
| Skewness | Intersection values higher than skewness threshold | SKE (B3) | ||
| Kurtosis | Intersection values higher than kurtosis threshold | KUR (B4) | ||
| 90 percentile | Intersection values higher than 90 percentile threshold | PER90 (B5) ⁎ |
Note: (1) B1, B2, B3, B4, B5 are short representation of each statistical method for visualization purposes. (2) * marked methods were demonstrated in figures in the main text.