Visualization of multiple ontology mapping
outputs
Sehyun Oh
January 15, 2026
Three thresholds for LM model
##
## Plotting 9 files
## Attribute levels: bodysite, disease, treatment_name
## Threshold levels: 0.7, 0.8, 0.9
## Shapes per Threshold : 0.7 = 16, 0.8 = 17, 0.9 = 15
##
## Accuracy Summary Statistics:
## variable1 variable2 top_k accuracy_pct
## 1 bodysite 0.7 Top-1 NA
## 2 bodysite 0.7 Top-3 NA
## 3 bodysite 0.7 Top-5 NA
## 4 bodysite 0.8 Top-1 NA
## 5 bodysite 0.8 Top-3 NA
## 6 bodysite 0.8 Top-5 NA
## 7 bodysite 0.9 Top-1 NA
## 8 bodysite 0.9 Top-3 NA
## 9 bodysite 0.9 Top-5 NA
## 10 disease 0.7 Top-1 NA
## 11 disease 0.7 Top-3 NA
## 12 disease 0.7 Top-5 NA
## 13 disease 0.8 Top-1 NA
## 14 disease 0.8 Top-3 NA
## 15 disease 0.8 Top-5 NA
## 16 disease 0.9 Top-1 NA
## 17 disease 0.9 Top-3 NA
## 18 disease 0.9 Top-5 NA
## 19 treatment_name 0.7 Top-1 64.71
## 20 treatment_name 0.7 Top-3 73.88
## 21 treatment_name 0.7 Top-5 77.16
## 22 treatment_name 0.8 Top-1 65.40
## 23 treatment_name 0.8 Top-3 75.95
## 24 treatment_name 0.8 Top-5 80.10
## 25 treatment_name 0.9 Top-1 79.76
## 26 treatment_name 0.9 Top-3 85.64
## 27 treatment_name 0.9 Top-5 87.89
##
##
## Accuracy Table (%):
## # A tibble: 9 × 5
## variable1 variable2 `Top-1` `Top-3` `Top-5`
## <fct> <fct> <dbl> <dbl> <dbl>
## 1 bodysite 0.7 NA NA NA
## 2 bodysite 0.8 NA NA NA
## 3 bodysite 0.9 NA NA NA
## 4 disease 0.7 NA NA NA
## 5 disease 0.8 NA NA NA
## 6 disease 0.9 NA NA NA
## 7 treatment_name 0.7 64.7 73.9 77.2
## 8 treatment_name 0.8 65.4 76.0 80.1
## 9 treatment_name 0.9 79.8 85.6 87.9
##
## Plot saved to: om_stage2_threshold/lm_3attr_3thresh.png
Three thresholds for ST model
##
## Plotting 9 files
## Attribute levels: bodysite, disease, treatment_name
## Threshold levels: 0.7, 0.8, 0.9
## Shapes per Threshold : 0.7 = 16, 0.8 = 17, 0.9 = 15
##
## Accuracy Summary Statistics:
## variable1 variable2 top_k accuracy_pct
## 1 bodysite 0.7 Top-1 NA
## 2 bodysite 0.7 Top-3 NA
## 3 bodysite 0.7 Top-5 NA
## 4 bodysite 0.8 Top-1 NA
## 5 bodysite 0.8 Top-3 NA
## 6 bodysite 0.8 Top-5 NA
## 7 bodysite 0.9 Top-1 NA
## 8 bodysite 0.9 Top-3 NA
## 9 bodysite 0.9 Top-5 NA
## 10 disease 0.7 Top-1 NA
## 11 disease 0.7 Top-3 NA
## 12 disease 0.7 Top-5 NA
## 13 disease 0.8 Top-1 NA
## 14 disease 0.8 Top-3 NA
## 15 disease 0.8 Top-5 NA
## 16 disease 0.9 Top-1 NA
## 17 disease 0.9 Top-3 NA
## 18 disease 0.9 Top-5 NA
## 19 treatment_name 0.7 Top-1 73.53
## 20 treatment_name 0.7 Top-3 78.55
## 21 treatment_name 0.7 Top-5 80.80
## 22 treatment_name 0.8 Top-1 80.10
## 23 treatment_name 0.8 Top-3 84.60
## 24 treatment_name 0.8 Top-5 87.02
## 25 treatment_name 0.9 Top-1 81.31
## 26 treatment_name 0.9 Top-3 85.29
## 27 treatment_name 0.9 Top-5 87.02
##
##
## Accuracy Table (%):
## # A tibble: 9 × 5
## variable1 variable2 `Top-1` `Top-3` `Top-5`
## <fct> <fct> <dbl> <dbl> <dbl>
## 1 bodysite 0.7 NA NA NA
## 2 bodysite 0.8 NA NA NA
## 3 bodysite 0.9 NA NA NA
## 4 disease 0.7 NA NA NA
## 5 disease 0.8 NA NA NA
## 6 disease 0.9 NA NA NA
## 7 treatment_name 0.7 73.5 78.6 80.8
## 8 treatment_name 0.8 80.1 84.6 87.0
## 9 treatment_name 0.9 81.3 85.3 87.0
##
## Plot saved to: om_stage2_threshold/st_3attr_3thresh.png
LM vs. ST
Compare the ontology mapping accuracy of three attributes with the
Stage2 threshold of 0.9.
##
## Plotting 6 files
## Attribute levels: bodysite, disease, treatment_name
## Model levels: LM, ST
## Shapes per Model : LM = 16, ST = 17
##
## Accuracy Summary Statistics:
## variable1 variable2 top_k accuracy_pct
## 1 bodysite LM Top-1 NA
## 2 bodysite LM Top-3 NA
## 3 bodysite LM Top-5 NA
## 4 bodysite ST Top-1 NA
## 5 bodysite ST Top-3 NA
## 6 bodysite ST Top-5 NA
## 7 disease LM Top-1 NA
## 8 disease LM Top-3 NA
## 9 disease LM Top-5 NA
## 10 disease ST Top-1 NA
## 11 disease ST Top-3 NA
## 12 disease ST Top-5 NA
## 13 treatment_name LM Top-1 79.76
## 14 treatment_name LM Top-3 85.64
## 15 treatment_name LM Top-5 87.89
## 16 treatment_name ST Top-1 81.31
## 17 treatment_name ST Top-3 85.29
## 18 treatment_name ST Top-5 87.02
##
##
## Accuracy Table (%):
## # A tibble: 6 × 5
## variable1 variable2 `Top-1` `Top-3` `Top-5`
## <fct> <fct> <dbl> <dbl> <dbl>
## 1 bodysite LM NA NA NA
## 2 bodysite ST NA NA NA
## 3 disease LM NA NA NA
## 4 disease ST NA NA NA
## 5 treatment_name LM 79.8 85.6 87.9
## 6 treatment_name ST 81.3 85.3 87.0
##
## Plot saved to: om_stage2_threshold/lm_st_3attr_1thresh.png
Best result
ST with Stage2 threshold of 0.9
##
## Plotting 3 files
## Attribute levels: bodysite, disease, treatment_name
## Model levels: ST
## Shapes per Model : ST = 16
##
## Accuracy Summary Statistics:
## variable1 variable2 top_k accuracy_pct
## 1 bodysite ST Top-1 NA
## 2 bodysite ST Top-3 NA
## 3 bodysite ST Top-5 NA
## 4 disease ST Top-1 NA
## 5 disease ST Top-3 NA
## 6 disease ST Top-5 NA
## 7 treatment_name ST Top-1 81.31
## 8 treatment_name ST Top-3 85.29
## 9 treatment_name ST Top-5 87.02
##
##
## Accuracy Table (%):
## # A tibble: 3 × 5
## variable1 variable2 `Top-1` `Top-3` `Top-5`
## <fct> <fct> <dbl> <dbl> <dbl>
## 1 bodysite ST NA NA NA
## 2 disease ST NA NA NA
## 3 treatment_name ST 81.3 85.3 87.0
##
## Plot saved to: om_stage2_threshold/st_3attr_1thresh.png
