Datasets:

ClarusC64
/

clinical-boundary-distance-v0.1

scenario_id stringlengths 9 9	oxygen_requirement int64 0 3	resp_rate int64 18 31	map int64 63 83	vasopressor_requirement stringclasses 4 values	lactate float64 1.1 5.2	urine_output int64 10 72	mental_status stringclasses 3 values	support_escalation_trend stringclasses 2 values	reserve_loss_trend stringclasses 4 values	label int64 0 2
train_001	0	18	82	none	1.2	70	baseline	stable	none	0
train_002	0	19	80	none	1.4	65	baseline	stable	none	0
train_003	1	20	78	none	1.6	60	baseline	stable	mild	0
train_004	0	18	81	low	1.5	62	baseline	stable	mild	0
train_005	1	21	76	low	1.8	55	baseline	stable	mild	0
train_006	1	23	74	low	2.2	45	baseline	rising	mild	1
train_007	1	24	72	low	2.5	40	mild_confusion	rising	moderate	1
train_008	2	24	71	low	2.7	38	baseline	rising	moderate	1
train_009	2	25	70	moderate	3	34	mild_confusion	rising	moderate	1
train_010	1	24	72	moderate	2.8	36	mild_confusion	stable	moderate	1
train_011	2	27	68	moderate	3.6	25	confused	rising	severe	2
train_012	3	29	66	moderate	4.1	20	confused	rising	severe	2
train_013	3	30	64	high	4.8	15	confused	rising	severe	2
train_014	2	28	67	high	4.3	18	confused	rising	severe	2
train_015	3	31	63	high	5.2	10	confused	rising	severe	2
train_016	0	18	83	none	1.1	72	baseline	stable	none	0
train_017	1	22	75	low	2.1	48	baseline	rising	mild	1
train_018	2	26	69	moderate	3.2	30	mild_confusion	rising	moderate	1
train_019	3	30	65	high	4.9	14	confused	rising	severe	2
train_020	2	27	68	high	4	22	confused	stable	severe	2

What this dataset does

This dataset tests whether a model can estimate how close a patient is to a clinical collapse boundary.

The task is not to identify whether collapse has already occurred.

The task is to classify the remaining margin before ordinary monitoring or support may no longer be enough.

Core stability idea

Clinical deterioration often appears first as margin loss.

A patient may still have acceptable observations while support needs rise and reserve falls.

This dataset tests whether a model can detect safe margin, narrowing margin, or critical boundary proximity.

Prediction target

The label column has three classes.

Label 0 means safe margin.

Label 1 means narrowing margin.

Label 2 means critical boundary proximity.

Row structure

Each row contains:

scenario_id
oxygen_requirement
resp_rate
map
vasopressor_requirement
lactate
urine_output
mental_status
support_escalation_trend
reserve_loss_trend
label

oxygen_requirement uses:

0 = room air or minimal support
1 = low oxygen requirement
2 = high oxygen requirement
3 = near respiratory boundary

vasopressor_requirement uses:

none
low
moderate
high

mental_status uses:

baseline
mild_confusion
confused

support_escalation_trend uses:

stable
rising

reserve_loss_trend uses:

none
mild
moderate
severe

Evaluation

Submissions must contain:

scenario_id,prediction
test_001,0
test_002,1
test_003,2

Run:

python scorer.py predictions.csv

Optional truth path:

python scorer.py predictions.csv data/test.csv

The scorer reports:

Accuracy
Macro precision
Macro recall
Macro F1
Confusion matrix
Structural Note

This dataset tests margin awareness.

It is designed to prevent shortcut logic based only on current vital signs or single support variables.

The intended reasoning requires integration across respiratory load, circulatory support, perfusion, urine output, cognition, support escalation, and reserve loss.

The dataset does not expose the hidden rationale behind each label.

License

MIT

Downloads last month: 48

Number of rows:

Total file size:

9.99 kB