model_utils.py

import numpy as np
import joblib
import warnings
import matplotlib.pyplot as plt
import matplotlib
import shap
import os
import tempfile
from config import MODEL_PATH, FEATURE_NAMES

warnings.filterwarnings('ignore')

matplotlib.use('Agg')

plt.rcParams['font.family'] = ['DejaVu Sans']
plt.rcParams['axes.unicode_minus'] = False

def calculate_derived_features(age, weight, height, neutrophil, lymphocyte, platelet):
    height_m = height / 100
    bmi = weight / (height_m ** 2)
    nlr = neutrophil / lymphocyte if lymphocyte > 0 else 0
    plr = platelet / lymphocyte if lymphocyte > 0 else 0
    return bmi, nlr, plr

def create_shap_plot(shap_values, feature_values, feature_names, prediction_proba):
    shap_vals = shap_values[0][:, 1]  # Shape: (18,) - SHAP values for class 1
        
    temp_file = tempfile.NamedTemporaryFile(delete=False, suffix='.png')
    temp_filename = temp_file.name
    temp_file.close()
    
    fig, ax = plt.subplots(figsize=(10, 12))
    
    sorted_indices = np.argsort(np.abs(shap_vals))
    sorted_shap_vals = shap_vals[sorted_indices]
    sorted_feature_names = [feature_names[i] for i in sorted_indices]
    sorted_feature_values = feature_values[sorted_indices]
    
    colors = ['red' if val > 0 else 'blue' for val in sorted_shap_vals]
    bars = ax.barh(range(len(sorted_shap_vals)), sorted_shap_vals, color=colors, alpha=0.7)
    
    ax.set_yticks(range(len(sorted_feature_names)))
    ax.set_yticklabels([f"{name} = {val:.2f}" for name, val in zip(sorted_feature_names, sorted_feature_values)])
    ax.set_xlabel('SHAP Value (Impact on Prediction)', fontsize=12)
    ax.set_title(f'Feature Impact Analysis\nComplication Risk: {prediction_proba[1]*100:.1f}%', 
                fontsize=14, pad=20)
    
    ax.axvline(x=0, color='black', linestyle='-', alpha=0.3)
    
    for i, (bar, val) in enumerate(zip(bars, sorted_shap_vals)):
        if val != 0:
            ax.text(val + (0.001 if val > 0 else -0.001), i, f'{val:.3f}', 
                    va='center', ha='left' if val > 0 else 'right', fontsize=9)
    
    ax.text(0.02, 0.98, 'Red: Increases risk\nBlue: Decreases risk', 
            transform=ax.transAxes, va='top', ha='left', 
            bbox=dict(boxstyle='round', facecolor='white', alpha=0.8))
    
    plt.tight_layout()
    plt.savefig(temp_filename, dpi=300, bbox_inches='tight', 
                facecolor='white', edgecolor='none')
    plt.close()
    
    return temp_filename

def get_shap_explainer_and_values(model, input_data):
        background_data = np.array([[
            28, 65, 162, 24.7, 2, 1, 0, 1, 28, 11.5, 34.0, 
            250, 8.5, 12.0, 6.0, 1.8, 3.33, 139
        ]])
        
        explainer = shap.KernelExplainer(model.predict_proba, background_data)
        shap_values = explainer.shap_values(input_data, nsamples=100)
        
        return shap_values

        
def predict_outcome(age, weight, height, gravidity, parity, h_abortion, 
                   living_child, gestational_age, hemoglobin, hematocrit, 
                   platelet, mpv, pdw, neutrophil, lymphocyte):
    model = get_model()
    
    if model is None:
        return "خطا: مدل بارگذاری نشده است", "", None
    
    try:
        bmi, nlr, plr = calculate_derived_features(age, weight, height, neutrophil, lymphocyte, platelet)
        
        input_data = np.array([[
            age, weight, height, bmi, gravidity, parity, h_abortion,
            living_child, gestational_age, hemoglobin, hematocrit, platelet,
            mpv, pdw, neutrophil, lymphocyte, nlr, plr
        ]])
        
        prediction_proba = model.predict_proba(input_data)[0]
        prediction = model.predict(input_data)[0]
        
        if prediction == 0:
            result = f"🟢 پیش‌بینی: سالم (احتمال سالم بودن: {prediction_proba[0]*100:.1f}%)"
            risk_level = "کم"
        else:
            result = f"🔴 پیش‌بینی: پرخطر (احتمال عوارض: {prediction_proba[1]*100:.1f}%)"
            risk_level = "بالا"
        
        detailed_report = f"""
📊 **گزارش تفصیلی پیش‌بینی**

**نتیجه کلی:** {result}

**سطح ریسک:** {risk_level}

**ویژگی‌های محاسبه شده:**
- BMI: {bmi:.2f}
- NLR (نسبت نوتروفیل به لنفوسیت): {nlr:.2f}
- PLR (نسبت پلاکت به لنفوسیت): {plr:.2f}

**احتمالات تفصیلی:**
- احتمال سالم بودن: {prediction_proba[0]*100:.1f}%
- احتمال بروز عوارض: {prediction_proba[1]*100:.1f}%

⚠️ **توجه:** این پیش‌بینی صرفاً جهت کمک به تشخیص است و نباید جایگزین نظر پزشک شود.
        """
        
        shap_values = get_shap_explainer_and_values(model, input_data)
        
        shap_plot_path = create_shap_plot(
            shap_values,
            input_data[0], 
            FEATURE_NAMES, 
            prediction_proba
        )

        return result, detailed_report, shap_plot_path
        
    except Exception as e:
        return f"خطا در پردازش: {str(e)}", "", None


model = None

def get_model():
    global model
    if model is None:
        try:
            model = joblib.load(MODEL_PATH)
            print("Model loaded successfully!")
            return model
        except Exception as e:
            print(f"Error loading model: {e}")
            return None
    return model

def cleanup_temp_files():
    try:
        temp_dir = tempfile.gettempdir()
        for filename in os.listdir(temp_dir):
            if filename.endswith('.png') and 'tmp' in filename:
                try:
                    os.remove(os.path.join(temp_dir, filename))
                except:
                    pass
    except:
        pass