Data Analytics

Healthcare Data Analysis Project: Hospital Readmission & Patient Flow Analysis

Intermediate Difficulty

1 Est. Time

0 items Components Needed

About This Project

Hospitals sit on huge amounts of operational data (admissions, discharges, diagnoses, department, length of stay) that's rarely analyzed beyond basic reporting. This project treats a hospital admissions dataset purely as a data analysis exercise (no prediction modeling) — the goal is to answer questions administrators and doctors actually care about:

Which departments have the highest patient load and longest average stay?
What's the 30-day readmission rate, and which diagnoses drive it most?
Are there seasonal spikes (e.g., respiratory illness in winter, dehydration/heatstroke in summer)?
Does age group correlate with length of stay or readmission risk?
Which day of week / time of day sees peak admissions (useful for staff scheduling)?

The output is a set of clear visualizations and a short insights report — the kind of analysis a hospital administrator, insurance analyst, or public health researcher would actually use for planning and resource allocation.

You can use a public dataset (e.g., Kaggle's "Hospital Readmissions" or "Healthcare Dataset") or plug in anonymized data if you're building this for a client.

Steps

Load and inspect data — check structure, missing values, data types (dates as strings, categorical codes).
Clean data — parse admission/discharge dates, handle missing diagnosis fields, standardize department names.
Derive key metrics — length of stay (discharge − admission), readmission flag (same patient admitted again within 30 days), age groups.
Univariate analysis — distribution of age, length of stay, department load.
Readmission analysis — overall rate, rate by department, rate by diagnosis, rate by age group.
Time-based analysis — monthly/seasonal admission trends, day-of-week patterns.
Cross-tab / correlation analysis — length of stay vs age, vs department, vs diagnosis category.
Summarize insights — 4–5 clear findings with supporting charts, framed as recommendations.

Components Required

Step-by-Step Instructions

Source Code

Arduino / ESP32 Sketch

# hospital_readmission_analysis.py
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

sns.set_style("whitegrid")
BRAND_DARK = "#163a6e"
BRAND_ORANGE = "#ff8a3d"

# ---------------------------------------------------------
# 1. Load dataset (replace with real hospital records)
# df = pd.read_csv("hospital_admissions.csv")
# ---------------------------------------------------------
np.random.seed(7)
n = 2000
departments = ["Cardiology", "Orthopedics", "General Medicine", "Pediatrics",
               "Neurology", "Respiratory"]
diagnoses = ["Hypertension", "Fracture", "Diabetes", "Pneumonia",
             "Asthma", "Migraine", "Viral Fever", "Cardiac Arrest"]

start_dates = pd.to_datetime("2024-01-01") + pd.to_timedelta(
    np.random.randint(0, 545, n), unit="D"
)
stay_length = np.random.gamma(shape=2.0, scale=2.0, size=n).astype(int) + 1

df = pd.DataFrame({
    "patient_id": np.random.randint(1000, 1500, n),  # allows repeat patients
    "department": np.random.choice(departments, n, p=[.18, .15, .22, .15, .12, .18]),
    "diagnosis": np.random.choice(diagnoses, n),
    "age": np.random.randint(1, 90, n),
    "admission_date": start_dates,
})
df["length_of_stay"] = stay_length
df["discharge_date"] = df["admission_date"] + pd.to_timedelta(df["length_of_stay"], unit="D")

# ---------------------------------------------------------
# 2. Derived fields
# ---------------------------------------------------------
df["age_group"] = pd.cut(
    df["age"], bins=[0, 12, 18, 40, 60, 90],
    labels=["Child (0-12)", "Teen (13-18)", "Adult (19-40)", "Middle-age (41-60)", "Senior (60+)"]
)
df["month"] = df["admission_date"].dt.to_period("M").astype(str)
df["day_of_week"] = df["admission_date"].dt.day_name()
df["season"] = df["admission_date"].dt.month % 12 // 3 + 1
season_map = {1: "Winter", 2: "Spring", 3: "Summer", 4: "Autumn"}
df["season"] = df["season"].map(season_map)

# ---------------------------------------------------------
# 3. Readmission flag: same patient, admitted again within 30 days
# ---------------------------------------------------------
df = df.sort_values(["patient_id", "admission_date"])
df["prev_discharge"] = df.groupby("patient_id")["discharge_date"].shift(1)
df["days_since_last_discharge"] = (df["admission_date"] - df["prev_discharge"]).dt.days
df["is_readmission"] = df["days_since_last_discharge"].between(0, 30)

overall_readmit_rate = df["is_readmission"].mean() * 100
print(f"Overall 30-day readmission rate: {overall_readmit_rate:.1f}%\n")

# ---------------------------------------------------------
# 4. Department load & avg length of stay
# ---------------------------------------------------------
dept_summary = df.groupby("department").agg(
    total_patients=("patient_id", "count"),
    avg_stay=("length_of_stay", "mean"),
    readmit_rate=("is_readmission", "mean")
).round(2).sort_values("total_patients", ascending=False)
dept_summary["readmit_rate"] = (dept_summary["readmit_rate"] * 100).round(1)
print("Department summary:\n", dept_summary, "\n")

plt.figure(figsize=(8, 5))
sns.barplot(data=dept_summary.reset_index(), x="total_patients", y="department", color=BRAND_DARK)
plt.title("Patient load by department")
plt.xlabel("Number of admissions")
plt.tight_layout()
plt.savefig("department_load.png", dpi=150)
plt.close()

# ---------------------------------------------------------
# 5. Readmission rate by diagnosis
# ---------------------------------------------------------
diag_readmit = df.groupby("diagnosis")["is_readmission"].mean().sort_values(ascending=False) * 100
plt.figure(figsize=(8, 5))
diag_readmit.plot(kind="barh", color=BRAND_ORANGE)
plt.title("30-day readmission rate by diagnosis (%)")
plt.xlabel("Readmission rate (%)")
plt.tight_layout()
plt.savefig("readmission_by_diagnosis.png", dpi=150)
plt.close()

# ---------------------------------------------------------
# 6. Seasonal admission trend
# ---------------------------------------------------------
seasonal = df.groupby(["season", "diagnosis"]).size().unstack(fill_value=0)
plt.figure(figsize=(10, 5))
seasonal.T.plot(kind="bar", stacked=True, colormap="tab20", ax=plt.gca())
plt.title("Diagnosis distribution by season")
plt.ylabel("Number of admissions")
plt.legend(title="Season", bbox_to_anchor=(1.02, 1), loc="upper left")
plt.tight_layout()
plt.savefig("seasonal_diagnosis_trend.png", dpi=150)
plt.close()

# ---------------------------------------------------------
# 7. Length of stay by age group
# ---------------------------------------------------------
plt.figure(figsize=(8, 5))
sns.boxplot(data=df, x="age_group", y="length_of_stay", color=BRAND_ORANGE)
plt.title("Length of stay by age group")
plt.xticks(rotation=20)
plt.tight_layout()
plt.savefig("stay_by_age_group.png", dpi=150)
plt.close()

# ---------------------------------------------------------
# 8. Admissions by day of week (staffing insight)
# ---------------------------------------------------------
day_order = ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"]
day_counts = df["day_of_week"].value_counts().reindex(day_order)
plt.figure(figsize=(8, 4))
day_counts.plot(kind="bar", color=BRAND_DARK)
plt.title("Admissions by day of week")
plt.ylabel("Number of admissions")
plt.xticks(rotation=30)
plt.tight_layout()
plt.savefig("admissions_by_weekday.png", dpi=150)
plt.close()

print("Charts saved: department_load.png, readmission_by_diagnosis.png,",
      "seasonal_diagnosis_trend.png, stay_by_age_group.png, admissions_by_weekday.png")

Completed this project?

Get an official Project Completion Certificate with a unique ID & QR verification — perfect for internships, resumes, and college submissions.

Get Certificate →