Table 2.

Results of internal and external validations.

	Data	Accuracy (95% CI)	RMSE	Pearson’s correlation coefficient
Internal validation	Cross-validation: healthy medical workers (N = 657)	97.0% (95.2%, 98.1%)	0.430	0.841
External validation	Mie National Hospital: healthy medical workers (input data for 3.5 weeks; N = 165)	98.2% (94.4%, 99.5%)	0.508	0.709
External validation	Mie National Hospital: healthy medical workers (input data for 12 weeks; N = 165)	94.5% (89.9%, 97.5%)	0.484	0.819
External validation	JSDT: hemodialysis patients (N = 192)	83.3% (77.1%, 88.2%)	0.924	0.721
External validation	JSDT: nondialysis elderly patients (N = 100)	83.8% (75.0%, 90.3%)	0.979	0.553

Accuracy = ((no. of patients whose 95% prediction interval included the actual value)/(no. of patients)) × 100 [%], and ${\mathrm{RMSE}}_t=\sqrt{\frac{1}{n}{\sum }_{i=1}^n{\left[\log Y_{it}-\log \left\{\mathrm{median}\left({\widehat{Y}}_{it}\right)\right\}\right]}^2}$. The RMSE and Pearson’s correlation coefficient between log Y_it and $\log \left\{\mathrm{median}\left({\widehat{Y}}_{it}\right)\right\}$ at 26 weeks were indicated for cross-validation, and RMSE and Pearson’s correlation coefficient at 24 weeks were indicated for Mie National Hospital and JSDT.

JSDT The Japanese Society for Dialysis Therapy.