Spaces:
Sleeping
Sleeping
| import gradio as gr | |
| import joblib | |
| import numpy as np | |
| import pandas as pd | |
| import warnings | |
| warnings.filterwarnings("ignore") | |
| import os | |
| # ── Load model ─────────────────────────────────────────────────────────────── | |
| MODEL_PATH = "Random_Forest.pth" | |
| model = joblib.load(MODEL_PATH) | |
| # ── Feature order must match training data ──────────────────────────────────── | |
| FEATURE_ORDER = [ | |
| "Sex", "Age", "Haemoglobin", "Hematocrit", "Lymphocytes", | |
| "MCH", "MCHC", "MCV", "Monocytes", "MPV", "PCT", "PDW", | |
| "Platelet Count", "Polymorphs", "RBC Count", "RDW SD", | |
| "WBC Count", "Eosinophils", "Creatinine", "Urea", "GFR " | |
| ] | |
| # ── Prediction function ─────────────────────────────────────────────────────── | |
| def predict_ckd( | |
| sex, age, haemoglobin, hematocrit, lymphocytes, | |
| mch, mchc, mcv, monocytes, mpv, pct, pdw, | |
| platelet_count, polymorphs, rbc_count, rdw_sd, | |
| wbc_count, eosinophils, creatinine, urea, gfr | |
| ): | |
| try: | |
| values = [ | |
| sex, age, haemoglobin, hematocrit, lymphocytes, | |
| mch, mchc, mcv, monocytes, mpv, pct, pdw, | |
| platelet_count, polymorphs, rbc_count, rdw_sd, | |
| wbc_count, eosinophils, creatinine, urea, gfr | |
| ] | |
| features = pd.DataFrame([values], columns=FEATURE_ORDER) | |
| pred = model.predict(features)[0] | |
| proba = model.predict_proba(features)[0] | |
| ckd_prob = round(float(proba[1]) * 100, 2) | |
| no_ckd_prob = round(float(proba[0]) * 100, 2) | |
| if pred == 1: | |
| label = "⚠️ CKD Detected" | |
| details = ( | |
| f"**Prediction: Chronic Kidney Disease (CKD)**\n\n" | |
| f"- CKD Probability: **{ckd_prob}%**\n" | |
| f"- No CKD Probability: {no_ckd_prob}%\n\n" | |
| f"> ⚠️ This result suggests the presence of Chronic Kidney Disease. " | |
| f"Please consult a qualified nephrologist for clinical evaluation." | |
| ) | |
| else: | |
| label = "✅ No CKD Detected" | |
| details = ( | |
| f"**Prediction: No Chronic Kidney Disease**\n\n" | |
| f"- No CKD Probability: **{no_ckd_prob}%**\n" | |
| f"- CKD Probability: {ckd_prob}%\n\n" | |
| f"> ✅ Based on the provided blood parameters, no CKD pattern was detected. " | |
| f"Regular health check-ups are still recommended." | |
| ) | |
| return label, details | |
| except Exception as e: | |
| import traceback | |
| err = traceback.format_exc() | |
| return "❌ Prediction Error", f"```\n{err}\n```" | |
| # ── UI ──────────────────────────────────────────────────────────────────────── | |
| DESCRIPTION = """ | |
| ## Chronic Kidney Disease (CKD) Prediction System | |
| **Research Project — PAF-IAST, HEC-Funded** | |
| This tool uses a **Random Forest classifier** trained on routine blood test parameters | |
| to assist in early screening of Chronic Kidney Disease. | |
| > **Disclaimer:** This application is intended for **research purposes only** and does | |
| > not constitute medical advice. Clinical decisions must always be made by a licensed physician. | |
| **Project Owner:** Dr. Muhammad Rizwan | |
| """ | |
| ARTICLE = """ | |
| ### About the Model | |
| The prediction model was developed as part of an HEC-funded research project at PAF-IAST. | |
| The Random Forest classifier was selected after a comprehensive comparison of 8 machine learning | |
| algorithms, including XGBoost, CatBoost, Gradient Boosting, and Logistic Regression. | |
| Explainability was ensured using SHAP (SHapley Additive exPlanations) and LIME analyses. | |
| Top predictive features include **Creatinine**, **GFR**, and **Urea** levels. | |
| """ | |
| with gr.Blocks( | |
| title="CKD Prediction | PAF-IAST", | |
| theme=gr.themes.Soft(primary_hue="blue", secondary_hue="indigo"), | |
| css=""" | |
| .result-box { border-radius: 10px; padding: 10px; } | |
| footer { display: none !important; } | |
| """ | |
| ) as demo: | |
| gr.Markdown(DESCRIPTION) | |
| with gr.Row(): | |
| # ── Column 1: Demographic + CBC ────────────────────────────────────── | |
| with gr.Column(): | |
| gr.Markdown("### 👤 Demographic") | |
| sex = gr.Radio( | |
| choices=[("Male", 1), ("Female", 0)], | |
| label="Sex", value=1 | |
| ) | |
| age = gr.Slider(minimum=1, maximum=100, step=1, value=49, label="Age (years)") | |
| gr.Markdown("### 🩸 Complete Blood Count") | |
| haemoglobin = gr.Number(value=11.0, label="Haemoglobin (g/dL)") | |
| hematocrit = gr.Number(value=32.0, label="Hematocrit (%)") | |
| rbc_count = gr.Number(value=3.7, label="RBC Count (millions/µL)") | |
| wbc_count = gr.Number(value=7.2, label="WBC Count (×10³/µL)") | |
| platelet_count = gr.Number(value=200, label="Platelet Count (×10³/µL)") | |
| # ── Column 2: Red Cell Indices + Differential ───────────────────────── | |
| with gr.Column(): | |
| gr.Markdown("### 📊 Red Cell Indices") | |
| mch = gr.Number(value=29.2, label="MCH (pg)") | |
| mchc = gr.Number(value=35.0, label="MCHC (g/dL)") | |
| mcv = gr.Number(value=85.0, label="MCV (fL)") | |
| rdw_sd = gr.Number(value=15.7, label="RDW-SD (fL)") | |
| mpv = gr.Number(value=8.7, label="MPV (fL)") | |
| pct = gr.Number(value=0.17, label="PCT (%)") | |
| pdw = gr.Number(value=37.0, label="PDW (%)") | |
| gr.Markdown("### 🔬 Differential Count (%)") | |
| lymphocytes = gr.Number(value=22, label="Lymphocytes (%)") | |
| monocytes = gr.Number(value=4, label="Monocytes (%)") | |
| polymorphs = gr.Number(value=71, label="Polymorphs (%)") | |
| eosinophils = gr.Number(value=2, label="Eosinophils (%)") | |
| # ── Column 3: Kidney Function ───────────────────────────────────────── | |
| with gr.Column(): | |
| gr.Markdown("### 🫘 Kidney Function Tests") | |
| creatinine = gr.Number(value=7.0, label="Creatinine (mg/dL)") | |
| urea = gr.Number(value=112.0, label="Blood Urea (mg/dL)") | |
| gfr = gr.Number(value=7.2, label="GFR (mL/min/1.73m²)") | |
| gr.Markdown("### 🔍 Prediction") | |
| predict_btn = gr.Button("🔮 Predict", variant="primary", size="lg") | |
| result_label = gr.Textbox(label="Result", interactive=False, lines=1) | |
| result_details = gr.Markdown(label="Details") | |
| gr.Markdown(ARTICLE) | |
| # ── Example inputs ──────────────────────────────────────────────────────── | |
| gr.Examples( | |
| examples=[ | |
| # CKD positive (high creatinine, low GFR) | |
| [1, 49, 9.3, 25.0, 12, 28.0, 36.0, 78.0, 5, 10.9, 0.08, 41.8, | |
| 76, 71, 3.5, 16.2, 5.5, 3, 13.2, 112, 4.06], | |
| # No CKD (normal values) | |
| [0, 36, 13.4, 40.0, 25, 31.0, 34.0, 90.0, 4, 8.5, 0.17, 36.0, | |
| 274, 68, 4.4, 12.9, 7.5, 2, 0.80, 37, 105.01], | |
| ], | |
| inputs=[ | |
| sex, age, haemoglobin, hematocrit, lymphocytes, | |
| mch, mchc, mcv, monocytes, mpv, pct, pdw, | |
| platelet_count, polymorphs, rbc_count, rdw_sd, | |
| wbc_count, eosinophils, creatinine, urea, gfr | |
| ], | |
| label="Sample Inputs (click to auto-fill)" | |
| ) | |
| predict_btn.click( | |
| fn=predict_ckd, | |
| inputs=[ | |
| sex, age, haemoglobin, hematocrit, lymphocytes, | |
| mch, mchc, mcv, monocytes, mpv, pct, pdw, | |
| platelet_count, polymorphs, rbc_count, rdw_sd, | |
| wbc_count, eosinophils, creatinine, urea, gfr | |
| ], | |
| outputs=[result_label, result_details] | |
| ) | |
| if __name__ == "__main__": | |
| demo.launch() |