Start with a CSV: Your Clinical Data Table

Before equations or machine learning algorithms, every clinical prediction study begins with a structured patient cohort stored in a row-and-column table.

Everyday Analogy: The Hospital Floor Census

Think of a CSV file as a morning hospital census chart. Each row is a unique patient in a bed. Each column is a specific clinical vital or lab recorded on admission. If a nurse didn't draw a blood glucose, that cell is empty (missing). You cannot begin treating the patient without understanding the census first.

Python Walkthrough

Here is how a clinical research team inspects an acute ischemic stroke cohort loaded from a CSV using Python's pandas library:

import pandas as pd

# 1. Load the admission stroke cohort
df = pd.read_csv("stroke_cohort.csv")

# 2. Check cohort size: (patients, clinical predictors)
print(df.shape)  # e.g., (626, 14) -> 626 patients, 14 variables

# 3. Inspect data types and missingness
print(df.dtypes)
print(df.isna().sum())

df.shape tells you your sample size immediately. In clinical modeling, sample size governs how many parameters your model can safely estimate without overfitting.

Medical Context: Stroke Predictors

In ischemic stroke thrombectomy studies, key baseline predictors include:

NIHSS (0–42): National Institutes of Health Stroke Scale (ordinal neurological severity).
Infarct Volume (mm³): Core ischemic volume estimated from CT Perfusion.
Admission NWU (%): Net Water Uptake measuring early ischemic brain edema.
Time Metrics (hours): Last Known Well (LKW) to hospital arrival.

Interactive Patient Cohort Table n=626 Patients Sample

Explore the admission stroke cohort. Missing values are flagged in orange.

ID	Age	Sex	NIHSS	Infarct (mL)	NWU (%)	mRS (90d)
#101	68	M	16	42.5	8.4%	2
#102	74	F	21	78.1	14.2%	4
#103	59	M	9	18.0	5.1%	1
#104	82	F	19	65.4	Missing	5
#105	63	M	14	34.2	7.9%	2

Key Clinical Takeaway: Before running regressions or ML algorithms, always inspect variable distributions, units, and missingness patterns.

Next Lesson → Lesson 2: Linear Regression