Skip to main content

Table 2 Mean Abs Error and RMSE in weeks in final machine learning model

From: Machine learning guided postnatal gestational age assessment using new-born screening metabolomic data in South Asia and sub-Saharan Africa

STATISTICS

Cohort

Africa

Asia

Overall

SGA

Overall

SGA

Overall

SGA

Training Dataset

Test Data Set- Pooled remaining

Test Data Set - remaining

Test Data Set- remaining

80% Pemba Samples + 80% Asian samples

20% PembaSamples +

20% Asiansamples

20% Pemba Samples

 

20% Asian Samples

 

MAE (95% CI)*

0.74 (0.65–0.98)

0.76 (0.65–0.88)

0.75 (0.61–0.89)

0.88 (0.75–1.16)

0.72 (0.62–0.88)

0.73 (0.61–0.95)

RMSE(95% CI)*

1.02 (0.91–1.14)

1.05 (0.91–1.19)

1.04 (0.89–1.16)

1.20 (1.10–1.31)

1.00 (0.89–1.16)

1.01 (0.93–1.19)

1 week difference (%)*

85.21 (72.31–94.65)

83.9 (71.21–92.32)

83.21 (78.31–90.05)

72 (65.67–79.34)

87.71 (76.63–95.39)

87.09 (77.67–94.21)

2 weeks difference (%)*

99.61 (91.42–100)

98.31 (89.74–100)

100 (93.32–100)

100 (92–79-100)

99.12 (91.56–100)

99.15 (90.45–100)

Training Dataset 80% Africa samples for Africa and 80% Asia samples for Asia

  

Test Dataset (20% Africa samples)

 

Test Dataset (20% Asia samples

 

MAE (95% CI)*

  

0.71 (0.58–0.85)

0.83 (0.71–1.10)

0.68 (0.58–0.87)

0.71 (0.62–0.83)

RMSE (95% CI)*

  

0.96 (0.82–1.07)

1.13 (1.01–1.27)

0.93 (0.82–01.05)

0.97 (0.84–1.08)

  1. *Bootstrapped,
  2. *Detailed description of the analytes used in the models have been given in supplementary information
Back to article page

BMC Pregnancy and Childbirth

ISSN: 1471-2393

Contact us