Datasets:

ClarusC64
/

clinical-constraint-pressure-v0.2

scenario_id stringlengths 9 9	oxygen_requirement int64 0 3	vasopressor_requirement stringclasses 4 values	renal_function stringclasses 4 values	urine_output stringclasses 4 values	mental_status stringclasses 3 values	respiratory_reserve stringclasses 4 values	circulatory_reserve stringclasses 4 values	renal_reserve stringclasses 4 values	support_trend stringclasses 3 values	reserve_trend stringclasses 3 values	label int64 0 2
train_001	0	none	normal	normal	baseline	high	high	high	stable	stable	0
train_002	1	none	mild_impairment	normal	baseline	medium	high	medium	stable	stable	0
train_003	1	low	mild_impairment	reduced	baseline	medium	medium	medium	stable	stable	1
train_004	1	low	mild_impairment	reduced	baseline	low	low	low	rising	worsening	2
train_005	2	low	moderate_impairment	reduced	baseline	medium	medium	medium	stable	stable	1
train_006	2	low	moderate_impairment	reduced	baseline	low	low	low	rising	worsening	2
train_007	2	moderate	moderate_impairment	oliguria	mild_confusion	low	low	low	rising	worsening	2
train_008	2	moderate	moderate_impairment	oliguria	mild_confusion	medium	medium	medium	stable	stable	1
train_009	3	moderate	severe_impairment	oliguria	confused	low	low	very_low	rising	worsening	2
train_010	3	moderate	severe_impairment	oliguria	confused	medium	medium	low	stable	stable	1
train_011	0	low	normal	normal	baseline	high	medium	high	stable	stable	0
train_012	0	low	normal	normal	baseline	medium	low	medium	rising	worsening	1
train_013	1	none	normal	normal	baseline	medium	high	high	stable	stable	0
train_014	1	none	normal	normal	baseline	low	medium	medium	rising	worsening	1
train_015	1	moderate	mild_impairment	reduced	mild_confusion	medium	low	medium	stable	stable	1
train_016	1	moderate	mild_impairment	reduced	mild_confusion	low	very_low	low	rising	worsening	2
train_017	2	high	moderate_impairment	oliguria	mild_confusion	low	low	low	rising	worsening	2
train_018	2	high	moderate_impairment	oliguria	mild_confusion	medium	medium	medium	stable	stable	1
train_019	3	high	severe_impairment	anuria	confused	very_low	very_low	very_low	rising	worsening	2
train_020	3	high	severe_impairment	anuria	confused	low	low	low	stable	stable	2
train_021	0	none	mild_impairment	normal	baseline	high	high	medium	stable	stable	0
train_022	0	none	mild_impairment	reduced	baseline	high	medium	medium	rising	worsening	1
train_023	1	low	moderate_impairment	reduced	baseline	medium	medium	low	stable	stable	1
train_024	1	low	moderate_impairment	reduced	mild_confusion	low	low	very_low	rising	worsening	2
train_025	2	none	normal	normal	baseline	medium	high	high	stable	stable	0
train_026	2	none	normal	normal	baseline	low	medium	medium	rising	worsening	1
train_027	2	moderate	severe_impairment	oliguria	mild_confusion	medium	medium	low	stable	stable	1
train_028	2	moderate	severe_impairment	oliguria	confused	low	low	very_low	rising	worsening	2
train_029	3	low	moderate_impairment	reduced	baseline	medium	medium	medium	stable	stable	1
train_030	3	low	moderate_impairment	reduced	mild_confusion	low	low	low	rising	worsening	2
train_031	0	none	normal	normal	baseline	high	high	high	falling	improving	0
train_032	1	low	mild_impairment	normal	baseline	medium	medium	medium	falling	improving	0
train_033	2	low	moderate_impairment	reduced	baseline	medium	medium	medium	falling	improving	1
train_034	2	moderate	moderate_impairment	oliguria	mild_confusion	medium	medium	medium	falling	improving	1
train_035	3	moderate	severe_impairment	oliguria	confused	low	low	low	falling	improving	1
train_036	1	moderate	moderate_impairment	reduced	mild_confusion	low	low	low	rising	worsening	2
train_037	0	low	mild_impairment	reduced	baseline	medium	low	low	rising	worsening	1
train_038	2	none	mild_impairment	normal	baseline	medium	high	medium	stable	stable	0
train_039	3	high	severe_impairment	oliguria	confused	low	low	very_low	rising	worsening	2
train_040	1	low	normal	normal	baseline	high	high	high	stable	stable	0

What this dataset does

This dataset tests whether a model can estimate clinical constraint pressure.

The task is not to identify current illness severity.

The task is to classify how much pressure the patient system is under relative to available reserve.

What changed in v0.2

v0.2 adds counterfactual and adversarial cases.

Some rows have the same oxygen requirement or vasopressor requirement but different reserve states.

Some high-looking cases have preserved reserve and improving trends.

Some medium-looking cases are actually high pressure because reserve is falling and support needs are rising.

This makes the task harder than v0.1.

Core stability idea

Constraint pressure is not the same as visible severity.

Pressure rises when support needs increase and reserve capacity falls.

A patient with moderate support needs may be high pressure if reserve is collapsing.

A patient with high support needs may be medium pressure if reserve is preserved and support needs are falling.

Correct classification requires reasoning across support level, reserve capacity, organ function, and direction of change.

Prediction target

The label column has three classes.

Label 0 means low constraint pressure.

Label 1 means medium constraint pressure.

Label 2 means high constraint pressure.

Row structure

Each row contains:

scenario_id
oxygen_requirement
vasopressor_requirement
renal_function
urine_output
mental_status
respiratory_reserve
circulatory_reserve
renal_reserve
support_trend
reserve_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

renal_function uses:

normal
mild_impairment
moderate_impairment
severe_impairment

urine_output uses:

normal
reduced
oliguria
anuria

mental_status uses:

baseline
mild_confusion
confused

reserve fields use:

high
medium
low
very_low

support_trend uses:

falling
stable
rising

reserve_trend uses:

improving
stable
worsening

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 benchmark contains counterfactual and adversarial cases designed to prevent shortcut learning from oxygen requirement, vasopressor requirement, or current illness severity.

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

The goal is to evaluate whether models can detect pressure building before collapse becomes obvious.

License

MIT

Downloads last month: 34

Number of rows:

Total file size:

14.4 kB