Datasets:

ClarusC64
/

clinical-quad-ae-signal-background-noise-reporting-lag-causality-bias-v0.1

trial_id string	site_id string	week int64	true_ae_rate_per_100py float64	background_event_rate_per_100py float64	reporting_lag_days int64	serious_ae_fraction float64	attribution_bias_0_1 float64	meddra_coding_variance float64	dsmb_attention_score float64	signal_strength_z float64	stop_signal_next_30d int64	label_stop_signal_next_30d int64
TRIAL_AEL03	S01	4	6.2	5.8	18	0.22	0.62	0.1	0.48	0.3	0	0
TRIAL_AEL03	S01	6	6.5	6	16	0.24	0.6	0.12	0.5	0.4	0	0
TRIAL_AEL03	S02	8	7	6.2	21	0.27	0.68	0.15	0.56	0.7	0	0
TRIAL_AEL03	S02	10	7.6	6.4	24	0.3	0.72	0.18	0.6	1.1	1	1
TRIAL_AEL03	S03	12	8.1	6.6	26	0.32	0.75	0.2	0.64	1.3	1	1
TRIAL_AEL04	S01	5	5.9	5.7	14	0.21	0.58	0.09	0.46	0.2	0	0
TRIAL_AEL04	S02	7	6.3	5.9	17	0.23	0.61	0.11	0.49	0.3	0	0
TRIAL_AEL04	S03	9	7.3	6.3	22	0.28	0.7	0.16	0.58	0.9	1	1
TRIAL_AEL04	S03	11	8	6.5	25	0.31	0.74	0.19	0.63	1.2	1	1
TRIAL_AEL04	S04	6	6.6	6.1	19	0.25	0.66	0.13	0.53	0.6	0	0

Clinical Quad AE Noise Lag Attribution Bias v0.1

Each row is a site week safety snapshot.

Core quad

AE signal rate
Background noise rate
Reporting lag
Attribution bias

Target

label_stop_signal_next_30d

Files

data/train.csv
data/tester.csv
scorer.py

Evaluation

Run model on data/tester.csv
Return predictions row aligned
Score with scorer.py

License

MIT