Datasets:
trial_id string | site_id string | patient_id string | day int64 | dose_mg int64 | egfr int64 | conmed_count int64 | high_risk_conmed int64 | time_on_drug_days int64 | drug_level_ng_ml int64 | alt int64 | ast int64 | ae_next_7d int64 | label_ae_next_7d int64 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
TRIAL_QS01 | S01 | P0001 | 1 | 25 | 102 | 0 | 0 | 1 | 135 | 21 | 19 | 0 | 0 |
TRIAL_QS01 | S01 | P0002 | 3 | 25 | 96 | 1 | 0 | 3 | 150 | 23 | 21 | 0 | 0 |
TRIAL_QS01 | S02 | P0003 | 5 | 50 | 88 | 1 | 0 | 5 | 210 | 28 | 25 | 0 | 0 |
TRIAL_QS01 | S02 | P0004 | 7 | 50 | 74 | 2 | 0 | 7 | 235 | 34 | 31 | 0 | 0 |
TRIAL_QS01 | S03 | P0005 | 9 | 75 | 66 | 3 | 1 | 9 | 305 | 46 | 42 | 1 | 1 |
TRIAL_QS01 | S03 | P0006 | 11 | 75 | 69 | 3 | 0 | 11 | 290 | 38 | 35 | 0 | 0 |
TRIAL_QS02 | S01 | P0007 | 2 | 100 | 61 | 4 | 1 | 2 | 360 | 55 | 51 | 0 | 0 |
TRIAL_QS02 | S01 | P0008 | 6 | 100 | 58 | 4 | 1 | 6 | 395 | 71 | 66 | 1 | 1 |
TRIAL_QS02 | S02 | P0009 | 8 | 100 | 54 | 5 | 1 | 8 | 410 | 78 | 73 | 1 | 1 |
TRIAL_QS02 | S02 | P0010 | 12 | 50 | 80 | 2 | 0 | 12 | 240 | 33 | 30 | 0 | 0 |
Clinical Quad Dose Renal ConMed Time Safety Drift v0.1
What this dataset is
You test whether a model can detect when a patient in a drug trial is entering a safety risk state.
Each row is a patient state snapshot.
Core quad coupling
Dose level
Renal function
Concomitant medication load
Time on treatment
The label asks
Will an adverse event occur in the next 7 days
Columns
trial_id
site_id
patient_id
day
dose_mg
egfr
conmed_count
high_risk_conmed
time_on_drug_days
drug_level_ng_ml
alt
ast
ae_next_7d
label_ae_next_7d
Target label
label_ae_next_7d
Files
data/train.csv
tester.py
scorer.py
How to test
Run tester.py
It loads data/train.csv
Creates dummy probability predictions
Calls scorer.py
Prints metrics
License
MIT
- Downloads last month
- 26