@hongmaple0820/med-scale-research-os 0.43.0

package/target-research-platform/src/analysis/stage4_business/business_plan.py

@@ -0,0 +1,816 @@
+"""
+商业计划书生成模块
+汇总前三阶段数据，生成投融资方案
+"""
+from typing import Optional
+from datetime import datetime
+
+
+class BusinessPlanGenerator:
+    """商业计划书生成器"""
+
+    def __init__(self):
+        self.plans = {}
+
+    def generate(
+        self,
+        target_name: str,
+        disease: Optional[str] = None,
+        stage1_data: Optional[dict] = None,
+        stage2_data: Optional[dict] = None,
+        stage3_data: Optional[dict] = None,
+        milestone_data: Optional[dict] = None,
+        company_info: Optional[dict] = None
+    ) -> dict:
+        """
+        生成商业计划书
+
+        Args:
+            target_name: 靶点名称
+            disease: 疾病名称
+            stage1_data: 阶段1数据（表达与预后）
+            stage2_data: 阶段2数据（筛选与排序）
+            stage3_data: 阶段3数据（深度调研）
+            milestone_data: 里程碑数据
+            company_info: 公司信息
+
+        Returns:
+            商业计划书
+        """
+        plan = {
+            "success": True,
+            "target": target_name,
+            "disease": disease,
+            "timestamp": datetime.now().isoformat(),
+            "sections": {}
+        }
+
+        # 1. 执行摘要
+        plan["sections"]["executive_summary"] = self._generate_executive_summary(
+            target_name, disease, stage1_data, stage2_data, stage3_data
+        )
+
+        # 2. 科学依据
+        plan["sections"]["scientific_rationale"] = self._generate_scientific_rationale(
+            target_name, disease, stage1_data, stage3_data
+        )
+
+        # 3. 市场机会
+        plan["sections"]["market_opportunity"] = self._generate_market_opportunity(
+            target_name, disease, stage3_data
+        )
+
+        # 4. 竞争优势
+        plan["sections"]["competitive_advantage"] = self._generate_competitive_advantage(
+            target_name, disease, stage3_data
+        )
+
+        # 5. 开发计划
+        plan["sections"]["development_plan"] = self._generate_development_plan(
+            target_name, disease, milestone_data
+        )
+
+        # 6. 财务预测
+        plan["sections"]["financial_projection"] = self._generate_financial_projection(
+            target_name, disease, milestone_data, stage3_data
+        )
+
+        # 7. 融资需求
+        plan["sections"]["funding_requirements"] = self._generate_funding_requirements(
+            target_name, disease, milestone_data
+        )
+
+        # 8. 风险评估
+        plan["sections"]["risk_assessment"] = self._generate_risk_assessment(
+            target_name, disease, stage3_data, milestone_data
+        )
+
+        # 9. 团队与资源
+        plan["sections"]["team_and_resources"] = self._generate_team_section(
+            company_info
+        )
+
+        # 10. 里程碑与退出策略
+        plan["sections"]["milestones_and_exit"] = self._generate_milestones_exit(
+            target_name, disease, milestone_data, stage3_data
+        )
+
+        self.plans[f"{target_name}_{disease or 'general'}"] = plan
+        return plan
+
+    def _generate_executive_summary(
+        self,
+        target_name: str,
+        disease: Optional[str],
+        stage1_data: Optional[dict],
+        stage2_data: Optional[dict],
+        stage3_data: Optional[dict]
+    ) -> dict:
+        """生成执行摘要"""
+        summary_parts = []
+
+        # 靶点概述
+        summary_parts.append(f"本项目聚焦于{target_name}靶点")
+        if disease:
+            summary_parts.append(f"用于治疗{disease}")
+
+        # 科学依据
+        if stage1_data:
+            summary1 = stage1_data.get("summary", {})
+            if summary1.get("key_findings"):
+                summary_parts.append(f"科学验证显示{summary1['key_findings'][0]}")
+
+        # 市场机会
+        if stage3_data:
+            summary3 = stage3_data.get("summary", {})
+            developability = summary3.get("developability", {})
+            if developability.get("level") == "high":
+                summary_parts.append("具有高开发潜力")
+
+        # 资金需求
+        if stage3_data:
+            assessment = stage3_data.get("stages", {}).get("assessment", {})
+            recommendations = assessment.get("rd_recommendations", [])
+            if recommendations:
+                summary_parts.append(f"建议: {recommendations[0].get('recommendation', '')}")
+
+        return {
+            "title": f"{target_name}{' - ' + disease if disease else ''} 项目商业计划书",
+            "summary": "。".join(summary_parts) + "。",
+            "key_highlights": self._extract_key_highlights(
+                target_name, disease, stage1_data, stage2_data, stage3_data
+            )
+        }
+
+    def _extract_key_highlights(
+        self,
+        target_name: str,
+        disease: Optional[str],
+        stage1_data: Optional[dict],
+        stage2_data: Optional[dict],
+        stage3_data: Optional[dict]
+    ) -> list:
+        """提取关键亮点"""
+        highlights = []
+
+        # 科学验证
+        if stage1_data:
+            summary1 = stage1_data.get("summary", {})
+            confidence = summary1.get("confidence_level", "")
+            if confidence in ["high", "medium"]:
+                highlights.append(f"科学验证置信度: {confidence}")
+
+        # 成药性
+        if stage2_data:
+            summary2 = stage2_data.get("summary", {})
+            top_candidates = summary2.get("top_candidates", [])
+            if top_candidates:
+                highlights.append(f"成药性评分: {top_candidates[0].get('score', 0):.2f}")
+
+        # 临床进展
+        if stage3_data:
+            clinical = stage3_data.get("stages", {}).get("clinical_pipeline", {}).get("data", {})
+            if clinical:
+                total_trials = clinical.get("total_trials", 0)
+                if total_trials > 0:
+                    highlights.append(f"全球{total_trials}个临床试验")
+
+        # 可开发性
+        if stage3_data:
+            developability = stage3_data.get("summary", {}).get("developability", {})
+            if developability.get("score", 0) >= 70:
+                highlights.append("高开发潜力")
+
+        return highlights
+
+    def _generate_scientific_rationale(
+        self,
+        target_name: str,
+        disease: Optional[str],
+        stage1_data: Optional[dict],
+        stage3_data: Optional[dict]
+    ) -> dict:
+        """生成科学依据"""
+        rationale = {
+            "target_biology": {},
+            "expression_data": {},
+            "clinical_evidence": {},
+            "literature_support": {}
+        }
+
+        # 靶点生物学
+        if stage3_data:
+            function_data = stage3_data.get("stages", {}).get("target_function", {}).get("data", {})
+            if function_data:
+                encyclopedia = function_data.get("encyclopedia", {})
+                rationale["target_biology"] = {
+                    "overview": encyclopedia.get("overview", ""),
+                    "key_features": encyclopedia.get("key_features", []),
+                    "research_status": encyclopedia.get("research_status", {})
+                }
+
+        # 表达数据
+        if stage1_data:
+            tissue_expr = stage1_data.get("stages", {}).get("tissue_expression", {})
+            if tissue_expr:
+                rationale["expression_data"] = {
+                    "gtex_available": tissue_expr.get("gtex", {}).get("success", False),
+                    "hpa_available": tissue_expr.get("hpa", {}).get("success", False),
+                    "key_findings": stage1_data.get("summary", {}).get("key_findings", [])
+                }
+
+        # 临床证据
+        if stage3_data:
+            clinical = stage3_data.get("stages", {}).get("clinical_pipeline", {}).get("data", {})
+            if clinical:
+                rationale["clinical_evidence"] = {
+                    "total_trials": clinical.get("total_trials", 0),
+                    "active_trials": clinical.get("active_trials", 0),
+                    "phase_distribution": clinical.get("phase_distribution", {})
+                }
+
+        # 文献支持
+        if stage3_data:
+            literature = stage3_data.get("stages", {}).get("literature", {}).get("data", {})
+            if literature:
+                rationale["literature_support"] = {
+                    "total_papers": literature.get("search_results", {}).get("total_count", 0),
+                    "recent_papers": literature.get("papers", [])[:5]
+                }
+
+        return rationale
+
+    def _generate_market_opportunity(
+        self,
+        target_name: str,
+        disease: Optional[str],
+        stage3_data: Optional[dict]
+    ) -> dict:
+        """生成市场机会分析"""
+        opportunity = {
+            "disease_burden": {},
+            "market_size": {},
+            "unmet_needs": {},
+            "growth_drivers": {}
+        }
+
+        # 从stage3获取市场数据
+        if stage3_data:
+            cl_data = stage3_data.get("stages", {}).get("competitive_landscape", {}).get("data", {})
+            sections = cl_data.get("sections", {})
+            market = sections.get("market_analysis", {})
+            if market:
+                opportunity["market_size"] = {
+                    "maturity": market.get("maturity", ""),
+                    "opportunity": market.get("opportunity", ""),
+                    "indicators": market.get("indicators", {})
+                }
+
+            # 竞争格局
+            competition = sections.get("competitive_analysis", {})
+            if competition:
+                opportunity["competition"] = {
+                    "intensity": competition.get("competition_intensity", ""),
+                    "top_players": competition.get("top_players", [])[:5]
+                }
+
+        # 未满足需求
+        opportunity["unmet_needs"] = {
+            "description": f"{disease or '该疾病'}领域存在未满足的临床需求" if disease else "存在未满足的临床需求",
+            "current_treatments": "需要进一步调研",
+            "differentiation": "需要明确差异化优势"
+        }
+
+        # 增长驱动因素
+        opportunity["growth_drivers"] = [
+            "患者人群增长",
+            "诊断率提升",
+            "新疗法需求"
+        ]
+
+        return opportunity
+
+    def _generate_competitive_advantage(
+        self,
+        target_name: str,
+        disease: Optional[str],
+        stage3_data: Optional[dict]
+    ) -> dict:
+        """生成竞争优势分析"""
+        advantage = {
+            "differentiation": {},
+            "barriers_to_entry": {},
+            "partnership_opportunities": {}
+        }
+
+        # 差异化
+        advantage["differentiation"] = {
+            "target_rationale": f"{target_name}靶点的独特优势",
+            "modality": "需要明确药物形式优势",
+            "patient_selection": "需要开发生物标志物"
+        }
+
+        # 进入壁垒
+        if stage3_data:
+            cl_data = stage3_data.get("stages", {}).get("competitive_landscape", {}).get("data", {})
+            sections = cl_data.get("sections", {})
+            competition = sections.get("competitive_analysis", {})
+            intensity = competition.get("competition_intensity", "")
+
+            if intensity in ["high", "very_high"]:
+                advantage["barriers_to_entry"]["competition"] = "竞争激烈，需要差异化策略"
+            else:
+                advantage["barriers_to_entry"]["competition"] = "竞争相对较少，先发优势明显"
+
+        # 合作机会
+        advantage["partnership_opportunities"] = {
+            "pharma_companies": "寻找大型药企合作",
+            "biotech": "考虑Biotech公司合作",
+            "academic": "学术机构合作开发"
+        }
+
+        return advantage
+
+    def _generate_development_plan(
+        self,
+        target_name: str,
+        disease: Optional[str],
+        milestone_data: Optional[dict]
+    ) -> dict:
+        """生成开发计划"""
+        plan = {
+            "strategy": {},
+            "timeline": {},
+            "key_activities": {}
+        }
+
+        if milestone_data:
+            milestones = milestone_data.get("milestones", [])
+
+            # 开发策略
+            plan["strategy"] = {
+                "approach": "阶段性开发",
+                "go_nogo_criteria": "基于每个阶段的数据决定是否推进",
+                "risk_mitigation": "早期验证降低后期风险"
+            }
+
+            # 时间线
+            plan["timeline"] = {
+                "total_duration": f"{milestone_data.get('total_duration_months', 0)}个月",
+                "estimated_completion": milestone_data.get("estimated_completion", ""),
+                "phases": [
+                    {
+                        "name": m.get("phase_name"),
+                        "duration": f"{m.get('duration_months')}个月",
+                        "start": m.get("start_date"),
+                        "end": m.get("end_date")
+                    }
+                    for m in milestones
+                ]
+            }
+
+            # 关键活动
+            for milestone in milestones[:3]:  # 前3个阶段
+                phase = milestone.get("phase")
+                plan["key_activities"][phase] = {
+                    "deliverables": milestone.get("deliverables", []),
+                    "key_risks": milestone.get("key_risks", [])
+                }
+
+        return plan
+
+    def _generate_financial_projection(
+        self,
+        target_name: str,
+        disease: Optional[str],
+        milestone_data: Optional[dict],
+        stage3_data: Optional[dict]
+    ) -> dict:
+        """生成财务预测"""
+        projection = {
+            "cost_structure": {},
+            "revenue_potential": {},
+            "key_metrics": {}
+        }
+
+        # 成本结构
+        if milestone_data:
+            milestones = milestone_data.get("milestones", [])
+            cost_by_phase = {}
+            for m in milestones:
+                phase = m.get("phase_name")
+                cost = m.get("cost_estimate", {}).get("typical", 0)
+                cost_by_phase[phase] = cost
+
+            projection["cost_structure"] = {
+                "total_rd_cost": f"${milestone_data.get('total_cost_estimate', 0)}M",
+                "by_phase": cost_by_phase
+            }
+
+        # 收入潜力（简化估算）
+        if stage3_data:
+            cl_data = stage3_data.get("stages", {}).get("competitive_landscape", {}).get("data", {})
+            sections = cl_data.get("sections", {})
+            market = sections.get("market_analysis", {})
+            if market:
+                projection["revenue_potential"] = {
+                    "market_opportunity": market.get("opportunity", ""),
+                    "market_maturity": market.get("maturity", ""),
+                    "note": "需要更详细的市场规模数据"
+                }
+
+        # 关键指标
+        projection["key_metrics"] = {
+            "npv": "需要现金流预测",
+            "irr": "需要收入预测",
+            "payback_period": "需要商业化时间线"
+        }
+
+        return projection
+
+    def _generate_funding_requirements(
+        self,
+        target_name: str,
+        disease: Optional[str],
+        milestone_data: Optional[dict]
+    ) -> dict:
+        """生成融资需求"""
+        funding = {
+            "total_requirement": {},
+            "funding_rounds": {},
+            "use_of_funds": {}
+        }
+
+        if milestone_data:
+            total_cost = milestone_data.get("total_cost_estimate", 0)
+
+            # 总融资需求
+            funding["total_requirement"] = {
+                "amount": f"${total_cost}M",
+                "timeline": f"{milestone_data.get('total_duration_months', 0)}个月",
+                "note": "基于标准开发时间线估算"
+            }
+
+            # 融资轮次
+            funding["funding_rounds"] = [
+                {
+                    "round": "种子轮/A轮",
+                    "amount": f"${total_cost * 0.15:.1f}M",
+                    "purpose": "靶点验证和Hit-to-Lead",
+                    "timeline": "0-24个月"
+                },
+                {
+                    "round": "B轮",
+                    "amount": f"${total_cost * 0.25:.1f}M",
+                    "purpose": "先导化合物优化和PCC",
+                    "timeline": "24-48个月"
+                },
+                {
+                    "round": "C轮",
+                    "amount": f"${total_cost * 0.35:.1f}M",
+                    "purpose": "IND申报和I期临床",
+                    "timeline": "48-72个月"
+                },
+                {
+                    "round": "D轮/IPO",
+                    "amount": f"${total_cost * 0.25:.1f}M",
+                    "purpose": "II/III期临床",
+                    "timeline": "72个月以后"
+                }
+            ]
+
+            # 资金用途
+            funding["use_of_funds"] = {
+                "rd": "60%",
+                "operations": "20%",
+                "regulatory": "10%",
+                "business_development": "10%"
+            }
+
+        return funding
+
+    def _generate_risk_assessment(
+        self,
+        target_name: str,
+        disease: Optional[str],
+        stage3_data: Optional[dict],
+        milestone_data: Optional[dict]
+    ) -> dict:
+        """生成风险评估"""
+        risks = {
+            "scientific_risks": [],
+            "regulatory_risks": [],
+            "commercial_risks": [],
+            "financial_risks": [],
+            "mitigation_strategies": []
+        }
+
+        # 科学风险
+        if stage3_data:
+            assessment = stage3_data.get("stages", {}).get("assessment", {})
+            pain_points = assessment.get("pain_points", [])
+            for pp in pain_points:
+                risks["scientific_risks"].append({
+                    "risk": pp.get("description", ""),
+                    "severity": pp.get("severity", "medium")
+                })
+
+        # 监管风险
+        risks["regulatory_risks"] = [
+            {"risk": "临床试验延迟", "severity": "medium"},
+            {"risk": "审评标准变化", "severity": "low"}
+        ]
+
+        # 商业风险
+        risks["commercial_risks"] = [
+            {"risk": "竞争产品上市", "severity": "medium"},
+            {"risk": "市场准入困难", "severity": "medium"}
+        ]
+
+        # 财务风险
+        risks["financial_risks"] = [
+            {"risk": "融资困难", "severity": "medium"},
+            {"risk": "成本超支", "severity": "medium"}
+        ]
+
+        # 缓解策略
+        risks["mitigation_strategies"] = [
+            "分阶段开发，降低早期风险",
+            "建立战略合作，分散风险",
+            "灵活调整开发策略",
+            "保持充足现金储备"
+        ]
+
+        return risks
+
+    def _generate_team_section(self, company_info: Optional[dict]) -> dict:
+        """生成团队与资源部分"""
+        if not company_info:
+            return {
+                "note": "需要提供公司信息",
+                "key_roles": [
+                    "首席科学官（CSO）",
+                    "首席医学官（CMO）",
+                    "研发副总裁",
+                    "商务拓展副总裁"
+                ],
+                "advisory_board": "需要组建科学顾问委员会"
+            }
+
+        return {
+            "company_name": company_info.get("name", ""),
+            "founded": company_info.get("founded", ""),
+            "team_size": company_info.get("team_size", ""),
+            "key_personnel": company_info.get("key_personnel", []),
+            "capabilities": company_info.get("capabilities", [])
+        }
+
+    def _generate_milestones_exit(
+        self,
+        target_name: str,
+        disease: Optional[str],
+        milestone_data: Optional[dict],
+        stage3_data: Optional[dict]
+    ) -> dict:
+        """生成里程碑与退出策略"""
+        result = {
+            "key_milestones": [],
+            "exit_strategies": [],
+            "partnership_options": []
+        }
+
+        # 关键里程碑
+        if milestone_data:
+            milestones = milestone_data.get("milestones", [])
+            for m in milestones[:5]:  # 前5个关键里程碑
+                result["key_milestones"].append({
+                    "phase": m.get("phase_name"),
+                    "target_date": m.get("end_date"),
+                    "deliverables": m.get("deliverables", [])[:2]
+                })
+
+        # 退出策略
+        result["exit_strategies"] = [
+            {
+                "strategy": "战略合作/授权",
+                "timing": "II期临床后",
+                "potential_value": "里程碑付款+ royalties"
+            },
+            {
+                "strategy": "并购",
+                "timing": "III期临床或上市后",
+                "potential_value": "基于管线价值"
+            },
+            {
+                "strategy": "IPO",
+                "timing": "II期临床数据后",
+                "potential_value": "市场估值"
+            }
+        ]
+
+        # 合作选项
+        result["partnership_options"] = [
+            {
+                "type": "共同开发",
+                "description": "与大型药企合作开发",
+                "pros": "分担风险和成本",
+                "cons": "分享收益"
+            },
+            {
+                "type": "授权-out",
+                "description": "将权益授权给合作伙伴",
+                "pros": "早期获得收入",
+                "cons": "失去控制权"
+            },
+            {
+                "type": "自主开发",
+                "description": "独立完成开发",
+                "pros": "保留全部权益",
+                "cons": "承担全部风险"
+            }
+        ]
+
+        return result
+
+    def to_markdown(self, plan: dict) -> str:
+        """转换为Markdown格式"""
+        if not plan.get("success"):
+            return "商业计划书生成失败"
+
+        sections = plan.get("sections", {})
+        target = plan.get("target", "")
+        disease = plan.get("disease", "")
+
+        md = f"""# {target}{' - ' + disease if disease else ''} 商业计划书
+
+**生成时间:** {plan.get('timestamp', '')}
+
+---
+
+## 1. 执行摘要
+
+{sections.get('executive_summary', {}).get('summary', '')}
+
+### 关键亮点
+
+"""
+        for highlight in sections.get('executive_summary', {}).get('key_highlights', []):
+            md += f"- {highlight}\n"
+
+        # 科学依据
+        scientific = sections.get('scientific_rationale', {})
+        md += f"""
+---
+
+## 2. 科学依据
+
+### 靶点生物学
+
+{scientific.get('target_biology', {}).get('overview', '')}
+
+### 表达数据
+
+"""
+        for finding in scientific.get('expression_data', {}).get('key_findings', []):
+            md += f"- {finding}\n"
+
+        # 市场机会
+        market = sections.get('market_opportunity', {})
+        md += f"""
+---
+
+## 3. 市场机会
+
+- **市场成熟度:** {market.get('market_size', {}).get('maturity', 'N/A')}
+- **市场机会:** {market.get('market_size', {}).get('opportunity', 'N/A')}
+
+### 未满足需求
+
+{market.get('unmet_needs', {}).get('description', '')}
+
+"""
+
+        # 开发计划
+        development = sections.get('development_plan', {})
+        timeline = development.get('timeline', {})
+        md += f"""
+---
+
+## 4. 开发计划
+
+- **总时长:** {timeline.get('total_duration', 'N/A')}
+- **预计完成:** {timeline.get('estimated_completion', 'N/A')}
+
+### 阶段计划
+
+| 阶段 | 时长 | 开始 | 结束 |
+|------|------|------|------|
+"""
+
+        for phase in timeline.get('phases', []):
+            md += f"| {phase.get('name', '')} | {phase.get('duration', '')} | {phase.get('start', '')} | {phase.get('end', '')} |\n"
+
+        # 融资需求
+        funding = sections.get('funding_requirements', {})
+        md += f"""
+---
+
+## 5. 融资需求
+
+- **总需求:** {funding.get('total_requirement', {}).get('amount', 'N/A')}
+- **时间线:** {funding.get('total_requirement', {}).get('timeline', 'N/A')}
+
+### 融资轮次
+
+| 轮次 | 金额 | 用途 | 时间 |
+|------|------|------|------|
+"""
+
+        for round_info in funding.get('funding_rounds', []):
+            md += f"| {round_info.get('round', '')} | {round_info.get('amount', '')} | {round_info.get('purpose', '')} | {round_info.get('timeline', '')} |\n"
+
+        # 风险评估
+        risks = sections.get('risk_assessment', {})
+        md += f"""
+---
+
+## 6. 风险评估
+
+### 科学风险
+
+"""
+        for risk in risks.get('scientific_risks', []):
+            md += f"- {risk.get('risk', '')} (严重程度: {risk.get('severity', '')})\n"
+
+        md += f"""
+### 缓解策略
+
+"""
+        for strategy in risks.get('mitigation_strategies', []):
+            md += f"- {strategy}\n"
+
+        # 退出策略
+        exit_section = sections.get('milestones_and_exit', {})
+        md += f"""
+---
+
+## 7. 退出策略
+
+"""
+        for exit in exit_section.get('exit_strategies', []):
+            md += f"- **{exit.get('strategy', '')}:** {exit.get('timing', '')} - {exit.get('potential_value', '')}\n"
+
+        md += f"""
+---
+
+*本商业计划书由靶点研究平台自动生成*
+"""
+
+        return md
+
+    def validate_plan(self, plan: dict) -> tuple[bool, list[str]]:
+        """验证计划书完整性"""
+        issues = []
+
+        if not plan.get("success"):
+            issues.append("计划书生成失败")
+            return False, issues
+
+        sections = plan.get("sections", {})
+
+        required_sections = [
+            "executive_summary",
+            "scientific_rationale",
+            "market_opportunity",
+            "development_plan",
+            "funding_requirements"
+        ]
+
+        for section in required_sections:
+            if not sections.get(section):
+                issues.append(f"缺少{section}部分")
+
+        return len(issues) == 0, issues
+
+    def get_summary(self, target_name: str, disease: Optional[str] = None) -> dict:
+        """获取计划书摘要"""
+        key = f"{target_name}_{disease or 'general'}"
+
+        if key not in self.plans:
+            return {
+                "target": target_name,
+                "disease": disease,
+                "status": "not_generated"
+            }
+
+        plan = self.plans[key]
+        sections = plan.get("sections", {})
+
+        return {
+            "target": target_name,
+            "disease": disease,
+            "status": "generated",
+            "sections_count": len(sections),
+            "funding_requirement": sections.get("funding_requirements", {}).get("total_requirement", {}).get("amount", ""),
+            "timeline": sections.get("development_plan", {}).get("timeline", {}).get("total_duration", "")
+        }