dashboard.py

import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
import pandas as pd
import numpy as np
from collections import Counter
from typing import List, Dict
import re

def generate_dashboard(data):
    """Legacy function - kept for backwards compatibility"""
    return generate_comprehensive_dashboard(data, "Other")


def extract_items_from_field(data: List[Dict], field_name: str) -> List[str]:
    """Extract and split items from semicolon-separated field"""
    items = []
    for row in data:
        value = row.get(field_name, "")
        if value and isinstance(value, str):
            # Split by semicolon and clean
            parts = [p.strip() for p in value.split(';') if p.strip()]
            items.extend(parts)
    return items


def generate_comprehensive_dashboard(
    data: List[Dict],
    interviewee_type: str
) -> plt.Figure:
    """
    Generate comprehensive dashboard with multiple visualizations
    """
    
    if not data or len(data) == 0:
        # Return empty figure with message
        fig, ax = plt.subplots(figsize=(10, 6))
        ax.text(0.5, 0.5, 'No data available for visualization',
                ha='center', va='center', fontsize=14)
        ax.axis('off')
        return fig
    
    df = pd.DataFrame(data)
    
    # Determine number of subplots based on interviewee type
    if interviewee_type == "HCP":
        fig = create_hcp_dashboard(df)
    elif interviewee_type == "Patient":
        fig = create_patient_dashboard(df)
    else:
        fig = create_general_dashboard(df)
    
    plt.tight_layout()
    return fig


def create_hcp_dashboard(df: pd.DataFrame) -> plt.Figure:
    """Create dashboard for HCP interviews"""
    
    fig, axes = plt.subplots(2, 2, figsize=(14, 10))
    fig.suptitle('Healthcare Professional Interview Analysis', fontsize=16, fontweight='bold')
    
    # 1. Quality Score Distribution
    ax1 = axes[0, 0]
    if 'Quality Score' in df.columns:
        quality_scores = pd.to_numeric(df['Quality Score'], errors='coerce').dropna()
        if len(quality_scores) > 0:
            ax1.hist(quality_scores, bins=10, color='#3498db', edgecolor='black', alpha=0.7)
            ax1.axvline(quality_scores.mean(), color='red', linestyle='--', 
                       label=f'Mean: {quality_scores.mean():.2f}')
            ax1.set_xlabel('Quality Score')
            ax1.set_ylabel('Frequency')
            ax1.set_title('Transcript Quality Distribution')
            ax1.legend()
            ax1.grid(axis='y', alpha=0.3)
    
    # 2. Top Diagnoses
    ax2 = axes[0, 1]
    if 'Diagnoses' in df.columns:
        diagnoses = extract_items_from_field(df.to_dict('records'), 'Diagnoses')
        if diagnoses:
            diagnosis_counts = Counter(diagnoses)
            top_diagnoses = dict(diagnosis_counts.most_common(8))
            
            if top_diagnoses:
                labels = list(top_diagnoses.keys())
                # Truncate long labels
                labels = [label[:30] + '...' if len(label) > 30 else label for label in labels]
                values = list(top_diagnoses.values())
                
                bars = ax2.barh(labels, values, color='#2ecc71', edgecolor='black')
                ax2.set_xlabel('Frequency')
                ax2.set_title('Most Common Diagnoses')
                ax2.invert_yaxis()
                
                # Add value labels
                for i, bar in enumerate(bars):
                    width = bar.get_width()
                    ax2.text(width, bar.get_y() + bar.get_height()/2, 
                            f' {int(width)}', ha='left', va='center', fontsize=9)
    
    # 3. Prescription Analysis
    ax3 = axes[1, 0]
    if 'Prescriptions' in df.columns:
        prescriptions = extract_items_from_field(df.to_dict('records'), 'Prescriptions')
        if prescriptions:
            rx_counts = Counter(prescriptions)
            top_rx = dict(rx_counts.most_common(8))
            
            if top_rx:
                labels = list(top_rx.keys())
                labels = [label[:30] + '...' if len(label) > 30 else label for label in labels]
                values = list(top_rx.values())
                
                bars = ax3.barh(labels, values, color='#e74c3c', edgecolor='black')
                ax3.set_xlabel('Frequency')
                ax3.set_title('Most Mentioned Prescriptions')
                ax3.invert_yaxis()
                
                for i, bar in enumerate(bars):
                    width = bar.get_width()
                    ax3.text(width, bar.get_y() + bar.get_height()/2,
                            f' {int(width)}', ha='left', va='center', fontsize=9)
    
    # 4. Word Count by Transcript
    ax4 = axes[1, 1]
    if 'Word Count' in df.columns and 'Transcript ID' in df.columns:
        word_counts = pd.to_numeric(df['Word Count'], errors='coerce').dropna()
        transcript_ids = df['Transcript ID'][:len(word_counts)]
        
        if len(word_counts) > 0:
            bars = ax4.bar(range(len(word_counts)), word_counts, color='#9b59b6', 
                          edgecolor='black', alpha=0.7)
            ax4.set_xlabel('Transcript')
            ax4.set_ylabel('Word Count')
            ax4.set_title('Interview Length by Transcript')
            ax4.set_xticks(range(len(word_counts)))
            ax4.set_xticklabels(transcript_ids, rotation=45, ha='right')
            ax4.grid(axis='y', alpha=0.3)
            
            # Add mean line
            ax4.axhline(word_counts.mean(), color='red', linestyle='--',
                       label=f'Average: {int(word_counts.mean())}')
            ax4.legend()
    
    return fig


def create_patient_dashboard(df: pd.DataFrame) -> plt.Figure:
    """Create dashboard for Patient interviews"""
    
    fig, axes = plt.subplots(2, 2, figsize=(14, 10))
    fig.suptitle('Patient Interview Analysis', fontsize=16, fontweight='bold')
    
    # 1. Quality Score Distribution
    ax1 = axes[0, 0]
    if 'Quality Score' in df.columns:
        quality_scores = pd.to_numeric(df['Quality Score'], errors='coerce').dropna()
        if len(quality_scores) > 0:
            ax1.hist(quality_scores, bins=10, color='#3498db', edgecolor='black', alpha=0.7)
            ax1.axvline(quality_scores.mean(), color='red', linestyle='--',
                       label=f'Mean: {quality_scores.mean():.2f}')
            ax1.set_xlabel('Quality Score')
            ax1.set_ylabel('Frequency')
            ax1.set_title('Transcript Quality Distribution')
            ax1.legend()
            ax1.grid(axis='y', alpha=0.3)
    
    # 2. Top Symptoms
    ax2 = axes[0, 1]
    if 'Primary Symptoms' in df.columns:
        symptoms = extract_items_from_field(df.to_dict('records'), 'Primary Symptoms')
        if symptoms:
            symptom_counts = Counter(symptoms)
            top_symptoms = dict(symptom_counts.most_common(8))
            
            if top_symptoms:
                labels = list(top_symptoms.keys())
                labels = [label[:30] + '...' if len(label) > 30 else label for label in labels]
                values = list(top_symptoms.values())
                
                bars = ax2.barh(labels, values, color='#e67e22', edgecolor='black')
                ax2.set_xlabel('Frequency')
                ax2.set_title('Most Common Symptoms')
                ax2.invert_yaxis()
                
                for i, bar in enumerate(bars):
                    width = bar.get_width()
                    ax2.text(width, bar.get_y() + bar.get_height()/2,
                            f' {int(width)}', ha='left', va='center', fontsize=9)
    
    # 3. Patient Concerns
    ax3 = axes[1, 0]
    if 'Main Concerns' in df.columns:
        concerns = extract_items_from_field(df.to_dict('records'), 'Main Concerns')
        if concerns:
            concern_counts = Counter(concerns)
            top_concerns = dict(concern_counts.most_common(6))
            
            if top_concerns:
                # Create word cloud style pie chart
                labels = list(top_concerns.keys())
                labels = [label[:25] + '...' if len(label) > 25 else label for label in labels]
                sizes = list(top_concerns.values())
                colors_list = ['#ff6b6b', '#4ecdc4', '#45b7d1', '#f9ca24', '#6c5ce7', '#a29bfe']
                
                ax3.pie(sizes, labels=labels, autopct='%1.1f%%', startangle=90,
                       colors=colors_list[:len(sizes)])
                ax3.set_title('Distribution of Patient Concerns')
    
    # 4. Side Effects
    ax4 = axes[1, 1]
    if 'Side Effects' in df.columns:
        side_effects = extract_items_from_field(df.to_dict('records'), 'Side Effects')
        if side_effects:
            se_counts = Counter(side_effects)
            top_se = dict(se_counts.most_common(6))
            
            if top_se:
                labels = list(top_se.keys())
                labels = [label[:30] + '...' if len(label) > 30 else label for label in labels]
                values = list(top_se.values())
                
                bars = ax4.barh(labels, values, color='#e74c3c', edgecolor='black')
                ax4.set_xlabel('Frequency')
                ax4.set_title('Reported Side Effects')
                ax4.invert_yaxis()
                
                for i, bar in enumerate(bars):
                    width = bar.get_width()
                    ax4.text(width, bar.get_y() + bar.get_height()/2,
                            f' {int(width)}', ha='left', va='center', fontsize=9)
        else:
            ax4.text(0.5, 0.5, 'No side effects reported',
                    ha='center', va='center', transform=ax4.transAxes, fontsize=12)
            ax4.axis('off')
    
    return fig


def create_general_dashboard(df: pd.DataFrame) -> plt.Figure:
    """Create general dashboard"""
    
    fig, axes = plt.subplots(2, 2, figsize=(14, 10))
    fig.suptitle('General Interview Analysis', fontsize=16, fontweight='bold')
    
    # 1. Quality Score Distribution
    ax1 = axes[0, 0]
    if 'Quality Score' in df.columns:
        quality_scores = pd.to_numeric(df['Quality Score'], errors='coerce').dropna()
        if len(quality_scores) > 0:
            ax1.hist(quality_scores, bins=10, color='#3498db', edgecolor='black', alpha=0.7)
            ax1.axvline(quality_scores.mean(), color='red', linestyle='--',
                       label=f'Mean: {quality_scores.mean():.2f}')
            ax1.set_xlabel('Quality Score')
            ax1.set_ylabel('Frequency')
            ax1.set_title('Transcript Quality Distribution')
            ax1.legend()
            ax1.grid(axis='y', alpha=0.3)
    
    # 2. Word Count Distribution
    ax2 = axes[0, 1]
    if 'Word Count' in df.columns:
        word_counts = pd.to_numeric(df['Word Count'], errors='coerce').dropna()
        if len(word_counts) > 0:
            ax2.hist(word_counts, bins=15, color='#2ecc71', edgecolor='black', alpha=0.7)
            ax2.set_xlabel('Word Count')
            ax2.set_ylabel('Frequency')
            ax2.set_title('Interview Length Distribution')
            ax2.grid(axis='y', alpha=0.3)
    
    # 3. Processing Summary
    ax3 = axes[1, 0]
    if 'Quality Score' in df.columns:
        quality_scores = pd.to_numeric(df['Quality Score'], errors='coerce').dropna()
        
        categories = ['Excellent\n(>0.8)', 'Good\n(0.6-0.8)', 'Fair\n(0.4-0.6)', 'Poor\n(<0.4)']
        counts = [
            sum(quality_scores > 0.8),
            sum((quality_scores >= 0.6) & (quality_scores <= 0.8)),
            sum((quality_scores >= 0.4) & (quality_scores < 0.6)),
            sum(quality_scores < 0.4)
        ]
        
        colors_list = ['#2ecc71', '#f39c12', '#e67e22', '#e74c3c']
        bars = ax3.bar(categories, counts, color=colors_list, edgecolor='black', alpha=0.7)
        ax3.set_ylabel('Number of Transcripts')
        ax3.set_title('Quality Score Categories')
        ax3.grid(axis='y', alpha=0.3)
        
        # Add value labels
        for bar in bars:
            height = bar.get_height()
            if height > 0:
                ax3.text(bar.get_x() + bar.get_width()/2., height,
                        f'{int(height)}', ha='center', va='bottom', fontsize=10)
    
    # 4. Summary Statistics Table
    ax4 = axes[1, 1]
    ax4.axis('off')
    
    stats_data = []
    if 'Transcript ID' in df.columns:
        stats_data.append(['Total Transcripts', str(len(df))])
    
    if 'Quality Score' in df.columns:
        quality_scores = pd.to_numeric(df['Quality Score'], errors='coerce').dropna()
        if len(quality_scores) > 0:
            stats_data.append(['Avg Quality Score', f"{quality_scores.mean():.2f}"])
            stats_data.append(['Min Quality Score', f"{quality_scores.min():.2f}"])
            stats_data.append(['Max Quality Score', f"{quality_scores.max():.2f}"])
    
    if 'Word Count' in df.columns:
        word_counts = pd.to_numeric(df['Word Count'], errors='coerce').dropna()
        if len(word_counts) > 0:
            stats_data.append(['Avg Word Count', f"{int(word_counts.mean()):,}"])
            stats_data.append(['Total Words', f"{int(word_counts.sum()):,}"])
    
    if stats_data:
        table = ax4.table(cellText=stats_data, cellLoc='left',
                         colWidths=[0.5, 0.3], loc='center',
                         colLabels=['Metric', 'Value'])
        table.auto_set_font_size(False)
        table.set_fontsize(11)
        table.scale(1, 2)
        
        # Style the table
        for i in range(len(stats_data) + 1):
            if i == 0:
                table[(i, 0)].set_facecolor('#34495e')
                table[(i, 1)].set_facecolor('#34495e')
                table[(i, 0)].set_text_props(weight='bold', color='white')
                table[(i, 1)].set_text_props(weight='bold', color='white')
            else:
                if i % 2 == 0:
                    table[(i, 0)].set_facecolor('#ecf0f1')
                    table[(i, 1)].set_facecolor('#ecf0f1')
        
        ax4.set_title('Summary Statistics', fontsize=12, fontweight='bold', pad=20)
    
    return fig