# AI将年轻健康成人心电图上的ST段抬高解读为急性心肌梗死，导致不必要的紧急导管检查

- **ID:** `medical/misinterpretation-of-ecg-st-elevation-for-benign-early-repolarization`
- **领域:** medical
- **类别:** data_error
- **错误码:** `ECG-STEMI-FALSE-POSITIVE-BER`
- **验证级别:** ai_generated
- **修复率:** 82%

## 根因

良性早期复极（BER）是一种正常变异，见于高达5%的人群，尤其是年轻男性，特征为心前导联凹面型ST段抬高；AI模型常缺乏区分其与ST段抬高型心肌梗死（STEMI）模式的特异性。

## 版本兼容性

| 版本 | 状态 | 引入 | 弃用 |
|------|------|------|------|
| GE Marquette 12SL v24 | active | — | — |
| Philips DXL v3.0 | active | — | — |
| CardioSoft v6.73 | active | — | — |

## 解决方案

1. ```
   Implement a two-step AI pipeline: first detect ST elevation, then apply a secondary model trained specifically to distinguish BER from STEMI using features like J-point notching, concave ST morphology, and absence of reciprocal changes.
   ```
2. ```
   Require the AI to output a confidence score and flag any ECG with ST elevation in patients under 40 without chest pain for manual physician review before activating cath lab.
   ```
3. ```
   Integrate a clinical decision rule: if the patient is under 40, has no cardiac risk factors, and has concave ST elevation in V2-V4 without reciprocal changes, recommend serial troponin and repeat ECG in 6 hours rather than immediate catheterization.
   ```

## 无效尝试

- **** — BER is a diagnosis of exclusion; without chest pain or troponin elevation, immediate catheterization is not indicated and carries procedural risks. (80% 失败率)
- **** — Automated interpretations have high false-positive rates for STEMI in young populations; physician expertise is needed. (70% 失败率)
- **** — BER typically shows ST elevation in V2-V4 without reciprocal depression, while STEMI often shows reciprocal changes in inferior leads. (60% 失败率)