大模型应用：企业员工 EAP 智能分析系统：混元大模型在心理评估中的应用解析.118

const submitAssessment = async () => {
    // 1. 验证所有必填题是否都已作答
    const unAnswered = questions
        .filter(q => q.required)
        .some(q => !answers[q.id]);

    if (unAnswered) {
        ElMessage.warning('请完成所有必填题目的作答');
        return;
    }

    submitting.value = true; // 开启加载状态

    try {
        // 2. 数据格式化：将答案从字符串转换为数字
        const answersData = {};
        Object.keys(answers).forEach(qid => {
            answersData[qid] = parseInt(answers[qid]);
        });

        let response;
        // 3. 尝试调用真实后端 API
        try {
            response = await axios.post('http://localhost:8000/api/assessment/analyze', {
                answers: answersData
            });
        } catch (apiError) {
            // 4. 降级策略：如果后端不可用，生成高质量的模拟数据
            console.log('后端API不可用，使用模拟数据');
            response = {
                data: {
                    assessment_id: 'demo_' + Date.now(),
                    risk_level: '低风险',
                    risk_description: '您的心理状态整体良好...',
                    summary: '根据您的答题情况...',
                    dimension_scores: { '工作压力': 75, '情绪状态': 82, /*...*/ },
                    dimensions: { '工作压力': '分析文本...', /*...*/ },
                    suggestions: { work: [...], life: [...], mental: [...] }
                }
            };
        }

        // 5. 更新结果状态
        Object.assign(assessmentResult, response.data);

        // 6. 切换视图到结果页
        currentPage.value = 'result';
        loadingResult.value = true;

        // 7. 模拟分析延迟，随后初始化图表
        setTimeout(() => {
            loadingResult.value = false;
            Vue.nextTick(() => {
                initDimensionChart(); // 关键：DOM渲染完成后才画图表
            });
        }, 100);

    } catch (error) {
        console.error('提交评估失败:', error);
        ElMessage.error('评估提交失败，请稍后重试');
    } finally {
        submitting.value = false; // 结束加载状态
    }
};

const initDimensionChart = () => {
    const chartDom = document.getElementById('dimensionChart');
    if (!chartDom) return; // 安全校验

    const myChart = echarts.init(chartDom);

    // 1. 动态提取数据
    const dimensionScores = assessmentResult.dimension_scores || {};
    const dimensions = Object.keys(dimensionScores);
    const scores = dimensions.map(d => dimensionScores[d]);
    const maxScores = Array(dimensions.length).fill(100); // 构造满分参考线

    if (dimensions.length === 0) return;

    // 2. 配置 ECharts 选项
    const option = {
        title: { text: 'EAP评估各维度得分对比', left: 'center' },
        tooltip: { trigger: 'axis' },
        legend: { data: ['您的得分', '满分'], bottom: 0 },
        grid: { containLabel: true }, // 自适应标签
        xAxis: {
            type: 'category',
            data: dimensions,
            axisLabel: { rotate: 15 } // 标签倾斜防止重叠
        },
        yAxis: { type: 'value', max: 100, min: 0 },
        series: [
            {
                name: '您的得分',
                type: 'bar',
                data: scores,
                itemStyle: { color: '#409eff' }, // 品牌色
                label: { show: true, position: 'top' }
            },
            {
                name: '满分',
                type: 'bar',
                data: maxScores,
                itemStyle: { color: '#e5e9f2' }, // 浅灰色背景条
                label: { show: true, position: 'top' }
            }
        ]
    };

    myChart.setOption(option);

    // 3. 响应式适配
    window.addEventListener('resize', () => {
        myChart.resize();
    });
};

const restartAssessment = () => {
    // 1. 切换回评估页面
    currentPage.value = 'assessment';
    
    // 2. 清空答题记录 (answers 是 reactive 对象，需逐个删除 key)
    Object.keys(answers).forEach(key => delete answers[key]);
    
    // 3. 清空之前的评估结果
    Object.keys(assessmentResult).forEach(key => delete assessmentResult[key]);
    
    // 注意：不需要手动重置图表，因为切回 assessment 页面后，
    // result 页面的 DOM 会被 v-if 销毁，下次进入时会重新 initDimensionChart
};

const downloadReport = async () => {
    try {
        const assessmentId = assessmentResult.assessment_id || 'demo_report';
        
        // 1. 请求二进制流 (blob)
        const response = await axios.get(
            `http://localhost:8000/api/assessment/report/${assessmentId}/download`, 
            { responseType: 'blob' } // 关键：告诉 axios 处理二进制数据
        );
        
        // 2. 创建临时下载链接
        const url = window.URL.createObjectURL(new Blob([response.data]));
        const link = document.createElement('a');
        link.href = url;
        link.setAttribute('download', `EAP评估报告_${new Date().getTime()}.pdf`);
        
        // 3. 触发点击并清理
        document.body.appendChild(link);
        link.click();
        document.body.removeChild(link);
        window.URL.revokeObjectURL(url); // 释放内存
        
        ElMessage.success('报告下载成功');
    } catch (error) {
        ElMessage.error('下载报告失败，请稍后重试');
    }
};

def process_assessment(self, answers: Dict[str, int]) -> Dict[str, Any]:
    """完整的评估处理流程"""
    # 1. 处理反向题（例如：q10 是反向计分，需转换）
    processed_answers = self.process_reverse_questions(answers)
    
    # 2. 计算各维度原始分并转换为百分制 (0-100)
    dimension_scores = self.calculate_dimension_scores(processed_answers)
    
    # 3. 基于常模数据计算标准分 (Z-Score)，用于横向对比
    standard_scores = self.calculate_standard_scores(dimension_scores)
    
    # 4. 根据 Z 分数判定风险等级 (低/中/高)
    risk_info = self.determine_risk_level(standard_scores)
    
    return {
        "raw_answers": answers,
        "processed_answers": processed_answers,
        "dimension_scores": dimension_scores,
        "standard_scores": standard_scores,
        "risk_info": risk_info
    }

def generate_assessment_report(self, answers: Dict[str, int], scoring_engine) -> Dict[str, Any]:
    """生成完整的评估报告"""
    # 1. 获取基础统计数据
    assessment_data = scoring_engine.process_assessment(answers)
    dimension_scores = assessment_data["dimension_scores"]
    risk_info = assessment_data["risk_info"]

    # 2. 构建结构化 Prompt (提示词工程)
    prompt = f"""请根据以下EAP评估数据，生成一份专业的心理分析报告（请以JSON格式返回）：

【评估数据】
风险等级：{risk_info['risk_level']}
平均Z分数：{risk_info['avg_z_score']}
各维度得分：
- 工作压力：{dimension_scores['工作压力']}分
- 情绪状态：{dimension_scores['情绪状态']}分
- ... (其他维度)

【返回格式要求】
请以JSON格式返回，包含以下字段：
{{
    "summary": "整体评估总结",
    "dimensions": {{ "工作压力": "分析...", ... }},
    "suggestions": {{ "work": [...], "life": [...], "mental": [...] }}
}}
请确保分析内容专业、人性化和可操作。"""

    # 3. 调用大模型 API
    llm_response = self.call_hunyuan_api(prompt)

    # 4. 解析响应或触发降级
    if llm_response:
        llm_report = self.parse_llm_response(llm_response, assessment_data)
    else:
        print("大模型API调用失败，使用备用分析方案")
        llm_report = self.generate_fallback_report(assessment_data)

    # 5. 整合最终报告
    full_report = {
        "assessment_id": f"eap_{int(time.time())}",
        "create_time": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
        "dimension_scores": assessment_data["dimension_scores"],
        "standard_scores": assessment_data["standard_scores"],
        **llm_report  # 合并大模型生成的文本内容
    }

    return full_report

def generate_fallback_report(self, assessment_data: Dict[str, Any]) -> Dict[str, Any]:
    """生成备用分析报告（当API调用失败时使用）"""
    dimension_scores = assessment_data["dimension_scores"]
    risk_info = assessment_data["risk_info"]

    # 1. 定义基于规则的模板库
    dimension_analysis = {
        "工作压力": {
            "high": f"您的工作压力得分为{dimension_scores['工作压力']}分，处于较高水平...建议合理规划...",
            "medium": f"您的工作压力得分为{dimension_scores['工作压力']}分，处于中等水平...",
            "low": f"您的工作压力得分为{dimension_scores['工作压力']}分，处于较低水平..."
        },
        # ... 其他维度类似模板
    }

    # 2. 根据分数动态选择模板内容
    dimensions = {}
    for dimension, score in dimension_scores.items():
        if score >= 80:
            dimensions[dimension] = dimension_analysis[dimension]["high"]
        elif score >= 60:
            dimensions[dimension] = dimension_analysis[dimension]["medium"]
        else:
            dimensions[dimension] = dimension_analysis[dimension]["low"]

    # 3. 生成通用建议
    suggestions = {
        "work": ["合理规划每日工作任务...", "每天预留复盘时间..."],
        "life": ["保证充足睡眠...", "每周进行体育锻炼..."],
        "mental": ["学习冥想技巧...", "培养兴趣爱好..."]
    }
    
    # 高风险特殊干预
    if risk_info["risk_level"] == "高风险":
        suggestions["mental"].append("建议尽快联系专业心理咨询师...")

    return {
        "summary": f"您的EAP评估整体风险等级为{risk_info['risk_level']}...",
        "dimensions": dimensions,
        "risk_level": risk_info["risk_level"],
        "risk_description": risk_info["risk_description"],
        "suggestions": suggestions
    }

def parse_llm_response(self, llm_response: str, assessment_data: Dict[str, Any]) -> Dict[str, Any]:
    """解析大模型返回的响应"""
    try:
        cleaned_response = llm_response.strip()

        # 1. 去除 Markdown 代码块标记 (大模型常返回 ```json ... ```)
        if cleaned_response.startswith("```json"):
            cleaned_response = cleaned_response[7:]
        elif cleaned_response.startswith("```"):
            cleaned_response = cleaned_response[3:]
        if cleaned_response.endswith("```"):
            cleaned_response = cleaned_response[:-3]
        cleaned_response = cleaned_response.strip()

        # 2. 尝试解析 JSON
        response = json.loads(cleaned_response)

        # 3. 数据完整性校验与默认值填充
        return {
            "summary": response.get("summary", ""),
            "dimensions": response.get("dimensions", {}),
            "risk_level": assessment_data["risk_info"]["risk_level"], # 强制使用计算出的风险等级，防止AI胡说
            "risk_description": assessment_data["risk_info"]["risk_description"],
            "suggestions": response.get("suggestions", {"work": [], "life": [], "mental": []})
        }

    except json.JSONDecodeError as e:
        # 4. 解析失败兜底：如果AI返回的不是合法JSON，直接切换到备用方案
        print(f"JSON解析失败: {e}")
        return self.generate_fallback_report(assessment_data)