meddra_coding

Repo for MedDRA coding results

Word Embeddings

BlazingText with Subword Models (BTSM) were built on MedDRA 4-22 & anonymized AE+MH data

1. 100 dimension model

2. 500 dimension model

3. 100 dimention model with artificial augmentation

4. 500 dimention model with artificial augmentation

MedDRA coding models

Methods	Loss	Accuracy	Precision	Recall	F1
m0 500D BTSM + LSTM+Attension	0.665	0.917	0.929	0.917	0.920
m1 500D BTSM + 2BiLSTM Attension	0.889	0.928	0.937	0.928	0.930
m2 500D BTSM + Attension +LSTM			0.940	0.924	0.930
m3 100D BTSM + Attension +Conv+biLSTM	0.610	0.935	0.944	0.935	0.935
m0 100D BTSM + 2BiLSTM+Attension	0.557	0.926	0.939	0.927	0.928

with MedDRA 4-22

1. 500D BTSM + LSTM+Attension

2. 500D BTSM + 2BiLSTM Attension

3. 500D BTSM + Attension+LSTM

4. 100D BTSM + Attension + 3 parallel conv +biLSTM +dense

Earlystop 'val_loss' min_delta=0 patientce 3

3. 100D BTSM + BiLSTM+Attension
4. 100D BTSM + LSTM+Dense+Attension
5. 500D BTSM + LSTM+Attension
6. 500D BTSM + Attension+LSTM
7. 500D BTSM + 2BiLSTM+Attension
8. 500D BTSM + LSTM+Dense+Attension

without MedDRA 4-22