在线演示 — ReHealth Core

STEP 01

🎯

预测Predict

输入患者数据Enter patient data

STEP 02

💊

干预Intervene

等待预测完成Waiting for prediction

STEP 03

⚖️

归因Attribute

等待干预完成Waiting for intervention

STEP 04

📄

结算Settle

等待归因完成Waiting for attribution

🎯 心血管风险预测🎯 Cardiovascular Risk Prediction

输入患者临床特征，模型实时输出风险评分和SHAP特征归因Enter patient clinical features for real-time risk scoring and SHAP attribution

年龄Age

性别Sex

BMI

收缩压 (mmHg)Systolic BP

总胆固醇 (mg/dL)Total Cholesterol

空腹血糖 (mg/dL)Fasting Glucose

吸烟状态Smoking Status

家族史Family History

预测结果Prediction Results

模型输出 · AUC 0.839 · 16维特征Model Output · AUC 0.839 · 16-dim Features

预测完成Prediction Complete

—

风险百分位Risk Percentile

<200ms

推理延迟Inference Latency

SHAP 特征重要性Feature Importance

← 输入患者数据后点击"运行风险预测"← Enter patient data and click "Run Risk Prediction"

💊 个性化干预方案💊 Personalized Intervention Plan

基于风险因子自动生成的针对性干预建议Auto-generated targeted intervention based on risk factors

干预轨迹记录Intervention Trajectory Log

结构化记录 · PSM归因的必要数据输入Structured logging · Required data input for PSM attribution

轨迹已记录Trajectory Logged

T=1

干预分配变量Treatment Assignment

—

干预项目数Intervention Items

12月mo

预设随访期Follow-up Period

—

记录时间Timestamp

✅ 干预分配变量（T=1）已生成
✅ 基线风险评分已关联
✅ 干预方案结构化记录完成
✅ 数据已准备好进入PSM归因流程 ✅ Treatment assignment (T=1) generated
✅ Baseline risk score linked
✅ Intervention plan structurally logged
✅ Data ready for PSM attribution pipeline

← 确认干预方案后生成轨迹记录← Confirm intervention plan to generate trajectory log

⚖️ PSM 因果归因⚖️ PSM Causal Attribution

倾向评分匹配 · 消除选择偏倚 · 量化因果效应Propensity Score Matching · Eliminate Selection Bias · Quantify Causal Effect

PSM将对干预组中的每位患者，在对照组中找到基线特征最相似的匹配对象，消除可观测的选择偏倚。匹配后的效果差异才具有因果解释力。 PSM finds the most similar control match for each treated patient, eliminating observable selection bias. Only the post-matching effect difference has causal interpretability.

248

干预组样本量Treatment Group N

1:1

匹配比例Matching Ratio

0.2σ

卡钳宽度Caliper Width

协变量数Covariates

归因结果Attribution Results

ATT · 置信区间 · SMD匹配质量ATT · Confidence Intervals · SMD Match Quality

归因完成Attribution Complete

ATT (Average Treatment Effect on Treated)

-0.087

95% CI: [-0.112, -0.062]

p < 0.001 ✓ 统计显著Statistically Significant

匹配质量 · SMD对比Matching Quality · SMD Comparison

变量Variable	匹配前Pre	匹配后Post	状态Status
年龄Age	0.31	0.04	✓
BMI	0.24	0.06	✓
收缩压SBP	0.19	0.05	✓
吸烟史Smoking	0.28	0.07	✓
基线风险Baseline Risk	0.35	0.03	✓

✅ 所有协变量匹配后 SMD < 0.1，匹配质量良好All covariates SMD < 0.1 post-match, good matching quality

← 点击"运行PSM因果归因"← Click "Run PSM Causal Attribution"

📄 可结算证据报告📄 Billable Evidence Report

符合 NMPA RWE 标准 · PDF + JSON 双格式输出NMPA RWE Compliant · PDF + JSON Dual Format Output

报告已生成Report Generated

PSM 因果归因证据报告Causal Attribution Evidence Report Level 2 · RWE

报告编号Report ID—

干预类型Intervention Type综合干预Combined

随访时长Follow-up12 个月months

干预组 NTreatment N248

匹配后 SMDPost-match SMD0.042 (<0.1 ✓)

主要结局 ATTPrimary Outcome ATT-0.087 (p<0.001)

95% CI[-0.112, -0.062]

次要结局Secondary收缩压 -4.3mmHgSBP -4.3mmHg

方法论标准MethodologyNMPA RWE 2020

下载 PDF 报告Download PDF

导出 JSONExport JSON

四步闭环完成4-Step Loop Complete

从预测到结算的完整因果链路Complete causal chain from prediction to settlement

闭环完成Loop Complete

🎯

预测 ✓Predict ✓

风险评分已输出，基线协变量已锁定Risk score output, baseline covariates locked

💊

干预 ✓Intervene ✓

干预方案已确认，轨迹已结构化记录Plan confirmed, trajectory structurally logged

⚖️

归因 ✓Attribute ✓

ATT = -0.087，p < 0.001，因果效应显著ATT = -0.087, p < 0.001, significant causal effect

📄

结算 ✓Settle ✓

可结算证据报告已生成，符合NMPA RWE标准Billable evidence report generated, NMPA RWE compliant

这就是预防结算的完整闭环This is the complete prevention settlement loop

四步因果链路完整，证据可被支付方接受Complete 4-step causal chain, evidence acceptable by payers

申请真实环境接入Request Real Access 了解方法论Read Methodology

← 生成报告后查看闭环总结← Generate report to see loop summary

体验 四步闭环 演示系统Experience the 4-Step Loop Demo

体验四步闭环演示系统Experience the 4-Step Loop Demo