test.py

import streamlit as st
import plotly.graph_objects as go
from datetime import datetime

# --- 嘗試匯入 FPDF (防呆機制) ---
try:
    from fpdf import FPDF
    HAS_FPDF = True
except ImportError:
    HAS_FPDF = False

# --- 設定頁面配置 ---
st.set_page_config(
    page_title="RadShield AI: Prostate",
    page_icon="🛡️",
    layout="wide",
    initial_sidebar_state="expanded"
)

# --- PDF 生成引擎 (保持英文以確保相容性) ---
if HAS_FPDF:
    class PDFReport(FPDF):
        def header(self):
            self.set_font('Arial', 'B', 15)
            self.cell(0, 10, 'RadShield AI - Clinical Decision Report', 0, 1, 'C')
            self.ln(5)

        def footer(self):
            self.set_y(-15)
            self.set_font('Arial', 'I', 8)
            self.cell(0, 10, f'Page {self.page_no()}', 0, 0, 'C')

    def create_pdf(patient_data, risk_result):
        pdf = PDFReport()
        pdf.add_page()
        pdf.set_font("Arial", size=12)

        # 1. Patient Info
        pdf.set_font("Arial", 'B', 12)
        pdf.cell(0, 10, "1. Patient Information", 0, 1)
        pdf.set_font("Arial", size=10)
        pdf.cell(0, 6, f"Name: {patient_data['name_en']} (ID: {patient_data['id']})", 0, 1)
        pdf.cell(0, 6, f"Age: {patient_data['age']} | Date: {datetime.now().strftime('%Y-%m-%d')}", 0, 1)
        pdf.ln(5)

        # 2. Risk Assessment
        pdf.set_font("Arial", 'B', 12)
        pdf.cell(0, 10, "2. Risk Stratification", 0, 1)
        pdf.set_font("Arial", size=10)
        
        # Acute
        if risk_result['acute_level'] == 'High':
            pdf.set_text_color(220, 53, 69) 
        else:
            pdf.set_text_color(0, 0, 0)
        pdf.cell(0, 6, f"Acute Toxicity Risk: {risk_result['acute_level']} (Score: {risk_result['acute_score']}/100)", 0, 1)
        
        # Chronic
        if risk_result['chronic_level'] == 'High':
            pdf.set_text_color(220, 53, 69) 
        else:
            pdf.set_text_color(0, 0, 0)
        pdf.cell(0, 6, f"Late Toxicity Risk: {risk_result['chronic_level']} (Score: {risk_result['chronic_score']}/100)", 0, 1)
        pdf.set_text_color(0, 0, 0) # Reset color
        pdf.ln(5)

        # 3. Drivers
        pdf.set_font("Arial", 'B', 12)
        pdf.cell(0, 10, "3. Key Risk Drivers (Concept 4)", 0, 1)
        pdf.set_font("Arial", size=10)
        
        if risk_result['drivers_en']:
            for driver in risk_result['drivers_en']:
                pdf.cell(0, 6, f"- {driver}", 0, 1)
        else:
            pdf.cell(0, 6, "- No significant risk drivers identified.", 0, 1)
        pdf.ln(5)

        # 4. Recommendations
        pdf.set_font("Arial", 'B', 12)
        pdf.cell(0, 10, "4. Optimization Plan (Concept 3)", 0, 1)
        pdf.set_font("Arial", size=10)
        
        if risk_result['recommendations_en']:
            for rec in risk_result['recommendations_en']:
                pdf.multi_cell(0, 6, f"- {rec}")
        else:
            pdf.cell(0, 6, "- Standard of Care (SoC) recommended.", 0, 1)
            
        pdf.ln(10)
        pdf.set_font("Arial", 'I', 8)
        pdf.cell(0, 6, "Disclaimer: This report is generated by a prototype AI system.", 0, 1)

        return pdf.output(dest='S').encode('latin-1')

# --- 風險計算核心邏輯 ---
def calculate_risk(patient_data):
    acute_score = 10
    chronic_score = 5
    drivers = []     
    drivers_en = []  
    recommendations = []    
    recommendations_en = [] 

    # 1. 臨床因子
    if patient_data['ibd']:
        acute_score += 25
        chronic_score += 30
        drivers.append('發炎性腸道疾病 (IBD)')
        drivers_en.append('Inflammatory Bowel Disease (IBD)')
    if patient_data['diabetes']:
        chronic_score += 10
    if patient_data['anticoagulants']:
        acute_score += 15
        chronic_score += 25
        drivers.append('使用抗凝血劑 (Bleeding Risk)')
        drivers_en.append('Anticoagulant Usage (High Bleeding Risk)')
    
    # 2. 基因組因子 (Concept 4)
    if patient_data['atm_mutation']:
        acute_score += 20
        chronic_score += 25
        drivers.append('ATM 基因突變 (Radiosensitivity)')
        drivers_en.append('ATM Gene Mutation (Radiosensitivity)')
    if patient_data['nbn_mutation']:
        acute_score += 15
        chronic_score += 20
        drivers.append('NBN 基因突變')
        drivers_en.append('NBN Gene Mutation')
    if patient_data['lig4_mutation']:
        acute_score += 30
        chronic_score += 40
        drivers.append('LIG4 基因表現異常 (DNA Repair Defect)')
        drivers_en.append('LIG4 Expression Abnormality (DNA Repair Defect)')

    # 3. 劑量學因子
    v70 = patient_data['rectum_v70']
    v50 = patient_data['rectum_v50']
    
    if v70 > 20:
        chronic_score += (v70 - 20) * 2
        drivers.append(f'Rectum V70 過高 ({v70}%)')
        drivers_en.append(f'High Rectum V70 ({v70}%)')
    if v50 > 50:
        acute_score += (v50 - 50) * 1.5

    acute_score = min(acute_score, 100)
    chronic_score = min(chronic_score, 100)

    def get_level(score):
        if score < 30: return 'Low', 'green'
        if score < 60: return 'Intermediate', 'orange'
        return 'High', 'red'

    acute_level, acute_color = get_level(acute_score)
    chronic_level, chronic_color = get_level(chronic_score)

    # 產生建議
    if chronic_level == 'High' or v70 > 20:
        recommendations.append("💡 建議使用直腸隔離凝膠 (SpaceOAR) 以增加物理距離 (Concept 1)")
        recommendations.append("💡 需重新優化治療計畫：限制 Rectum V70 < 15%")
        recommendations_en.append("Apply Rectal Spacer (SpaceOAR) to increase physical separation (Concept 1).")
        recommendations_en.append("Re-optimize treatment plan: Tighten Rectum V70 constraint to < 15%.")
    
    if patient_data['atm_mutation'] or patient_data['lig4_mutation']:
        recommendations.append("🧬 基因組顯示高輻射敏感性：考慮減少單次劑量或轉用質子治療 (Proton)")
        recommendations_en.append("Genomic Radiosensitivity: Consider hypofractionation reduction or Proton Therapy.")
    
    if patient_data['anticoagulants'] and acute_score > 40:
        recommendations.append("💊 出血風險高：建議會診心臟科評估暫停抗凝血劑")
        recommendations_en.append("High Bleeding Risk: Consult cardiology regarding temporary suspension of anticoagulants.")

    return {
        "acute_score": acute_score,
        "chronic_score": chronic_score,
        "acute_level": acute_level,
        "chronic_level": chronic_level,
        "drivers": drivers,
        "drivers_en": drivers_en,
        "recommendations": recommendations,
        "recommendations_en": recommendations_en
    }

# --- 側邊欄導航與輸入 ---
with st.sidebar:
    st.title("🛡️ RadShield AI")
    st.caption("Prostate Cancer Precision RT")
    
    st.header("Patient Intake")
    
    with st.expander("1. Clinical Profile", expanded=True):
        p_name = st.text_input("Patient Name", "陳大明 (Chen, Da-Ming)")
        p_name_en = "Chen, Da-Ming" if "Chen" in p_name else "Patient-001" 
        p_id = st.text_input("Patient ID", "P-2024-089")
        p_age = st.number_input("Age", 40, 90, 68)
        
        st.markdown("---")
        st.subheader("Risk Factors")
        c_diabetes = st.checkbox("糖尿病 (Diabetes)")
        c_ibd = st.checkbox("發炎性腸病 (IBD)")
        c_anticoag = st.checkbox("使用抗凝血劑 (Blood Thinners)")
        c_smoking = st.checkbox("吸菸史 (Smoking History)")

    with st.expander("2. Radiogenomics Panel", expanded=True):
        st.info("Concept 4: Pre-therapy Gene Markers")
        g_atm = st.checkbox("ATM Mutation", help="與 DNA 雙股斷裂修復相關")
        g_nbn = st.checkbox("NBN Mutation", help="細胞週期檢查點")
        g_lig4 = st.checkbox("LIG4 Mutation", help="NHEJ 修復機制異常")

    with st.expander("3. Dosimetry (DVH)", expanded=True):
        st.warning("Concept 3: DVH Constraints")
        d_v70 = st.slider("Rectum V70 (%)", 0, 40, 18, help="Target < 15-20%")
        d_v50 = st.slider("Rectum V50 (%)", 0, 80, 35)

# --- 整合資料 ---
patient_data = {
    'name': p_name,
    'name_en': p_name_en,
    'id': p_id,
    'ibd': c_ibd,
    'diabetes': c_diabetes,
    'anticoagulants': c_anticoag,
    'atm_mutation': g_atm,
    'nbn_mutation': g_nbn,
    'lig4_mutation': g_lig4,
    'rectum_v70': d_v70,
    'rectum_v50': d_v50,
    'age': p_age
}

# --- 執行風險運算 ---
risk_result = calculate_risk(patient_data)

# --- 主畫面 UI ---
st.title(f"Patient: {p_name}")
st.markdown(f"**ID:** {p_id} | **Date:** {datetime.now().strftime('%Y-%m-%d')}")

tab1, tab2, tab3 = st.tabs(["📊 Risk Dashboard", "🧬 Analysis Detail", "📄 Decision Report"])

# Tab 1: Dashboard
with tab1:
    col1, col2 = st.columns(2)
    with col1:
        st.subheader("急性毒性風險 (Acute)")
        st.caption("治療期間 ~ 3個月內")
        delta_color = "normal" if risk_result['acute_level'] == 'Low' else "inverse"
        st.metric("Risk Score / 100", risk_result['acute_score'], risk_result['acute_level'], delta_color=delta_color)
        st.progress(risk_result['acute_score'] / 100)
    with col2:
        st.subheader("慢性毒性風險 (Late)")
        st.caption("治療後 6個月以上")
        delta_color_chronic = "normal" if risk_result['chronic_level'] == 'Low' else "inverse"
        st.metric("Risk Score / 100", risk_result['chronic_score'], risk_result['chronic_level'], delta_color=delta_color_chronic)
        st.progress(risk_result['chronic_score'] / 100)

    st.markdown("---")
    st.subheader("💡 Clinical Decision Support (CDS)")
    if risk_result['recommendations']:
        for rec in risk_result['recommendations']:
            if "SpaceOAR" in rec or "V70" in rec:
                st.error(rec)
            else:
                st.warning(rec)
    else:
        st.success("✅ 目前評估為低風險，建議維持標準治療計畫。")

# Tab 2: Analysis
with tab2:
    col_chart, col_text = st.columns([2, 1])
    with col_chart:
        categories = ['Dosimetry', 'Genomics', 'Comorbidities', 'Meds', 'Age']
        values = [
            min(d_v70 * 3, 100),
            90 if (g_atm or g_lig4) else 20,
            80 if (c_ibd or c_diabetes) else 20,
            80 if c_anticoag else 10,
            70 if p_age > 75 else 40
        ]
        fig = go.Figure()
        fig.add_trace(go.Scatterpolar(r=values, theta=categories, fill='toself', name='Risk Profile', line_color='#3b82f6'))
        fig.update_layout(polar=dict(radialaxis=dict(visible=True, range=[0, 100])), showlegend=False, title="Risk Radar")
        st.plotly_chart(fig, use_container_width=True)
    
    with col_text:
        st.subheader("主要風險因子")
        if risk_result['drivers']:
            for driver in risk_result['drivers']:
                st.write(f"🔴 **{driver}**")
        else:
            st.write("無顯著風險因子")

# Tab 3: Report (Modified)
with tab3:
    st.header("Clinical Report Preview")
    
    # 1. 準備 Markdown 格式的報告內容 (中文版，用於網頁顯示)
    report_markdown = f"""
    # RadShield AI - Patient Risk Report
    **Patient:** {patient_data['name']} ({patient_data['id']})
    **Date:** {datetime.now().strftime('%Y-%m-%d')}
    
    ## Risk Assessment
    - **Acute Toxicity Risk:** {risk_result['acute_level']} (Score: {risk_result['acute_score']})
    - **Late Toxicity Risk:** {risk_result['chronic_level']} (Score: {risk_result['chronic_score']})
    
    ## Identified Risk Drivers
    {chr(10).join(['- ' + d for d in risk_result['drivers']] if risk_result['drivers'] else ['- No significant drivers'])}
    
    ## Optimization Recommendations
    {chr(10).join(['- ' + r for r in risk_result['recommendations']] if risk_result['recommendations'] else ['- Maintain Standard of Care'])}
    
    ---
    *Generated by RadShield AI Prototype*
    """
    
    # 2. 在網頁上渲染 Markdown
    st.markdown(report_markdown)
    st.markdown("---")
    
    # 3. PDF 下載按鈕
    st.subheader("Download")
    if HAS_FPDF:
        if st.button("Generate PDF Report"):
            pdf_bytes = create_pdf(patient_data, risk_result)
            st.download_button(
                label="📥 Download PDF to EMR",
                data=pdf_bytes,
                file_name=f"RadShield_Report_{p_id}.pdf",
                mime="application/pdf"
            )
    else:
        st.warning("⚠️ 未安裝 'fpdf' 套件，僅提供純文字下載。")
        st.download_button(
            label="📥 Download Text Report",
            data=report_markdown, # 使用上面的 markdown 字串作為 txt 內容
            file_name=f"RadShield_Report_{p_id}.txt",
            mime="text/plain"
        )