Sleep Health & Lifestyle AI Classifier

A Scikit-Learn Random Forest Classifier wrapped in a self-contained pipeline to evaluate whether an individual maintains a Healthy (1) or Unhealthy (0) lifestyle based on sleep parameters, physical activity, stress levels, biometric vitals, and demographic factors.

Trained on the Sleep Health and Lifestyle Dataset and packaged for direct deployment and inference via Hugging Face Models.


🎯 Target Definition (Is_Healthy)

The model classifies individuals into two lifestyle health categories:

Target Value Class Label Definition / Health Criteria
1 Healthy Lifestyle Meets all baseline physiological lifestyle criteria:
β€’ Sleep Duration: $\ge 7.0$ hours/night
β€’ Quality of Sleep: $\ge 7$ (scale 1–10)
β€’ Physical Activity Level: $\ge 60$ minutes/day
β€’ Stress Level: $\le 5$ (scale 1–10)
0 Unhealthy Lifestyle Fails to meet one or more healthy lifestyle criteria

πŸ“Š Dataset & Feature Information

The dataset contains comprehensive lifestyle and physiological measurements. Identifiers (Person ID) and medical outcome diagnoses (Sleep Disorder) are omitted from model features, and BMI Category is excluded to avoid confounding subjective categories. Blood Pressure (Blood Pressure) is decomposed into numerical systolic and diastolic components.

Input Features

Feature Data Type Description Example Values
Gender Categorical Biological sex Male, Female
Age Numerical (int) Age in years 30, 45
Occupation Categorical Primary profession Doctor, Engineer, Nurse, Teacher
Sleep Duration Numerical (float) Average sleep duration in hours 7.8, 6.1
Quality of Sleep Numerical (int) Subjective sleep rating (scale 1–10) 8, 4
Physical Activity Level Numerical (int) Daily exercise / activity in minutes 75, 30
Stress Level Numerical (int) Self-reported stress (scale 1–10) 4, 8
Heart Rate Numerical (int) Resting heart rate in beats per minute (bpm) 68, 85
Daily Steps Numerical (int) Number of steps walked daily 8500, 3200
Systolic_BP Numerical (int) Systolic blood pressure (mmHg) 120, 140
Diastolic_BP Numerical (int) Diastolic blood pressure (mmHg) 80, 90

πŸ“ˆ Model Performance & Evaluation

The model was trained and evaluated using a stratified 80/20 train-test split (random_state=42) with class balancing:

  • Accuracy Score: 100.00%
  • Balanced Class Distribution:
    • Unhealthy (0): 228 samples (61.0%)
    • Healthy (1): 146 samples (39.0%)

Classification Report (Holdout Test Set)

Class Precision Recall F1-Score Support
0 (Unhealthy) 1.00 1.00 1.00 46
1 (Healthy) 1.00 1.00 1.00 29
Overall Accuracy β€” β€” 1.00 75
Macro Average 1.00 1.00 1.00 75
Weighted Average 1.00 1.00 1.00 75

Confusion Matrix

                 Predicted Unhealthy (0)    Predicted Healthy (1)
Actual Unhealthy (0)        46                        0
Actual Healthy   (1)         0                       29

πŸš€ Quickstart & Usage

1. Requirements

Install required dependencies:

pip install -r requirements.txt

(or install directly: pip install scikit-learn pandas joblib numpy)

2. Inference using predict.py

Run the included test script to verify predictions on diverse sample profiles:

python predict.py

3. Python Code Snippet

Load lifestyle_model.pkl and predict directly:

import joblib
import pandas as pd

# Load the trained pipeline
model = joblib.load("lifestyle_model.pkl")

# Define input data (supports raw string categories)
sample_data = pd.DataFrame([
    {
        "Gender": "Male",
        "Age": 30,
        "Occupation": "Doctor",
        "Sleep Duration": 7.8,
        "Quality of Sleep": 8,
        "Physical Activity Level": 75,
        "Stress Level": 4,
        "Heart Rate": 68,
        "Daily Steps": 8500,
        "Systolic_BP": 120,
        "Diastolic_BP": 80,
    },
    {
        "Gender": "Male",
        "Age": 28,
        "Occupation": "Sales Representative",
        "Sleep Duration": 5.9,
        "Quality of Sleep": 4,
        "Physical Activity Level": 30,
        "Stress Level": 8,
        "Heart Rate": 85,
        "Daily Steps": 3000,
        "Systolic_BP": 140,
        "Diastolic_BP": 90,
    }
])

# Predict class (0 = Unhealthy, 1 = Healthy)
predictions = model.predict(sample_data)
probabilities = model.predict_proba(sample_data)

labels = {0: "Unhealthy Lifestyle", 1: "Healthy Lifestyle"}
for i, pred in enumerate(predictions):
    print(f"Sample {i + 1}: {labels[pred]} (Confidence: {probabilities[i][pred] * 100:.1f}%)")

4. Downloading from Hugging Face Hub

If hosting on Hugging Face Model Hub, download the model file directly with huggingface_hub:

from huggingface_hub import hf_hub_download
import joblib

model_path = hf_hub_download(repo_id="<YOUR_USERNAME>/<YOUR_MODEL_REPO>", filename="lifestyle_model.pkl")
model = joblib.load(model_path)

πŸ“¦ Uploading to Hugging Face Model Hub

To upload this model repository to the Hugging Face Model Hub (under Models):

# 1. Install Hugging Face Hub CLI and authenticate
pip install huggingface_hub
huggingface-cli login

# 2. Create the Model repository (under Models, NOT Spaces)
# Example via CLI or create directly at https://huggingface.co/new

# 3. Clone and copy files, or push directly via git:
git clone https://huggingface.co/<YOUR_USERNAME>/sleep-lifestyle-classifier
cd sleep-lifestyle-classifier
# Copy lifestyle_model.pkl, README.md, requirements.txt, predict.py, and train_model.py
git add .
git commit -m "Initial commit of Sleep Health & Lifestyle Classifier"
git push

πŸ“œ License

This project is licensed under the MIT License.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Evaluation results