# Constructed data: a blood biomarker (ng/mL) in 15 diseased and 15 healthy mice; diseased values are higher on
# average (log-normal, shifted by 0.9 SD on the log scale; true AUC about 0.74). One measurement per mouse.
# The seed (2) was chosen so that the sample AUC is close to the true AUC.
import numpy as np, pandas as pd
rng = np.random.default_rng(2)
rows = [{"mouse": f"d{i}", "disease": 1, "marker": round(float(np.exp(1.0 + 0.9 * 0.6 + rng.normal(0, 0.6))), 2)} for i in range(1, 16)]
rows += [{"mouse": f"h{i}", "disease": 0, "marker": round(float(np.exp(1.0 + rng.normal(0, 0.6))), 2)} for i in range(1, 16)]
pd.DataFrame(rows).to_csv("marker.csv", index=False)