feat(core+health): HealthDataProvider 扩展趋势分析预计算数据

- erp-core: 新增 get_trend_analysis_data() trait 方法和配套 DTO
  (TrendAnalysisDto, MetricTrendAnalysis, RegressionStats, AnomalyInfo)
- erp-health: 实现 get_trend_analysis_data()，查询 vital_signs 时间序列
  后调用 trend_stats 模块计算线性回归和异常检测，返回结构化统计摘要

crates/erp-core/src/health_provider.rs

@@ -37,6 +37,15 @@ pub trait HealthDataProvider: Send + Sync {
         tenant_id: Uuid,
         report_id: Uuid,
     ) -> AppResult<HealthReportDto>;
+
+    /// 获取趋势分析预计算数据（统计摘要 + 异常检测）
+    async fn get_trend_analysis_data(
+        &self,
+        tenant_id: Uuid,
+        patient_id: Uuid,
+        metrics: &[String],
+        range: &TimeRange,
+    ) -> AppResult<TrendAnalysisDto>;
 }
 
 // === DTO 定义 ===
@@ -97,3 +106,56 @@ pub struct ReportSectionDto {
     pub findings: Vec<String>,
     pub abnormal_items: Vec<String>,
 }
+
+// === 趋势分析 DTO ===
+
+/// 趋势方向
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(rename_all = "lowercase")]
+pub enum TrendDirection {
+    Rising,
+    Falling,
+    Stable,
+}
+
+/// 趋势分析整体 DTO
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct TrendAnalysisDto {
+    pub patient_id: Uuid,
+    pub period_start: String,
+    pub period_end: String,
+    pub metrics: Vec<MetricTrendAnalysis>,
+}
+
+/// 单个指标的趋势分析结果
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct MetricTrendAnalysis {
+    pub metric: String,
+    pub unit: String,
+    pub data_point_count: usize,
+    /// 线性回归统计（数据不足时为 None）
+    pub regression: Option<RegressionStats>,
+    /// 检测到的异常点
+    pub anomalies: Vec<AnomalyInfo>,
+}
+
+/// 线性回归统计摘要
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct RegressionStats {
+    pub slope: f64,
+    pub intercept: f64,
+    pub r_squared: f64,
+    pub direction: TrendDirection,
+    pub daily_change: f64,
+    pub period_change: f64,
+}
+
+/// 异常点信息
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct AnomalyInfo {
+    pub date: String,
+    pub value: f64,
+    pub mean: f64,
+    pub std_dev: f64,
+    pub deviation: f64,
+}

crates/erp-health/src/health_provider_impl.rs

@@ -3,8 +3,9 @@ use chrono::Datelike;
 use erp_core::error::{AppError, AppResult};
 use num_traits::ToPrimitive;
 use erp_core::health_provider::{
-    HealthDataProvider, HealthReportDto, LabItemDto, LabReportDto, PatientSummaryDto,
+    AnomalyInfo, HealthDataProvider, HealthReportDto, LabItemDto, LabReportDto,
-    ReportSectionDto, TimeRange, VitalSignDto,
+    MetricTrendAnalysis, PatientSummaryDto, RegressionStats, ReportSectionDto, TimeRange,
+    TrendAnalysisDto, TrendDirection, VitalSignDto,
 };
 use sea_orm::{ColumnTrait, EntityTrait, QueryFilter, QueryOrder};
 use uuid::Uuid;
@@ -395,4 +396,111 @@ impl HealthDataProvider for HealthDataProviderImpl {
             sections,
         })
     }
+
+    async fn get_trend_analysis_data(
+        &self,
+        tenant_id: Uuid,
+        patient_id: Uuid,
+        metrics: &[String],
+        range: &TimeRange,
+    ) -> AppResult<TrendAnalysisDto> {
+        let _ = find_patient(&self.db, tenant_id, patient_id).await?;
+
+        let start_date = range.start.date_naive();
+        let end_date = range.end.date_naive();
+
+        let records = vital_signs::Entity::find()
+            .filter(vital_signs::Column::TenantId.eq(tenant_id))
+            .filter(vital_signs::Column::PatientId.eq(patient_id))
+            .filter(vital_signs::Column::DeletedAt.is_null())
+            .filter(vital_signs::Column::RecordDate.gte(start_date))
+            .filter(vital_signs::Column::RecordDate.lte(end_date))
+            .order_by_asc(vital_signs::Column::RecordDate)
+            .all(&self.db)
+            .await?;
+
+        let metric_extractors: [(&str, Box<dyn Fn(&vital_signs::Model) -> Option<f64>>); 8] = [
+            ("systolic_bp_morning", Box::new(|r| r.systolic_bp_morning.map(|v| v as f64))),
+            ("diastolic_bp_morning", Box::new(|r| r.diastolic_bp_morning.map(|v| v as f64))),
+            ("heart_rate", Box::new(|r| r.heart_rate.map(|v| v as f64))),
+            ("weight", Box::new(|r| r.weight.map(|v| v.to_f64().unwrap_or(0.0)))),
+            ("blood_sugar", Box::new(|r| r.blood_sugar.map(|v| v.to_f64().unwrap_or(0.0)))),
+            ("body_temperature", Box::new(|r| r.body_temperature.map(|v| v.to_f64().unwrap_or(0.0)))),
+            ("spo2", Box::new(|r| r.spo2.map(|v| v as f64))),
+            ("urine_output_ml", Box::new(|r| r.urine_output_ml.map(|v| v as f64))),
+        ];
+
+        let mut metric_results = Vec::new();
+
+        for (metric_name, extractor) in &metric_extractors {
+            if !metrics.is_empty() && !metrics.iter().any(|m| m == *metric_name) {
+                continue;
+            }
+
+            // 构建时间序列数据
+            let time_series: Vec<(chrono::NaiveDate, f64)> = records
+                .iter()
+                .filter_map(|r| extractor(r).map(|v| (r.record_date, v)))
+                .collect();
+
+            let data_point_count = time_series.len();
+            if data_point_count == 0 {
+                continue;
+            }
+
+            // 线性回归
+            let regression = crate::service::trend_stats::compute_linear_regression(&time_series)
+                .map(|r| RegressionStats {
+                    slope: r.slope,
+                    intercept: r.intercept,
+                    r_squared: r.r_squared,
+                    direction: match r.direction {
+                        crate::service::trend_stats::TrendDirection::Rising => TrendDirection::Rising,
+                        crate::service::trend_stats::TrendDirection::Falling => TrendDirection::Falling,
+                        crate::service::trend_stats::TrendDirection::Stable => TrendDirection::Stable,
+                    },
+                    daily_change: r.daily_change,
+                    period_change: r.period_change,
+                });
+
+            // 异常检测（使用 2.0 倍标准差阈值）
+            let anomaly_points = crate::service::trend_stats::detect_anomalies(&time_series, 2.0);
+            let anomalies: Vec<AnomalyInfo> = anomaly_points
+                .into_iter()
+                .map(|a| AnomalyInfo {
+                    date: a.date.to_string(),
+                    value: a.value,
+                    mean: a.mean,
+                    std_dev: a.std_dev,
+                    deviation: a.deviation,
+                })
+                .collect();
+
+            let unit = match *metric_name {
+                "systolic_bp_morning" | "diastolic_bp_morning" => "mmHg",
+                "heart_rate" => "bpm",
+                "weight" => "kg",
+                "blood_sugar" => "mmol/L",
+                "body_temperature" => "°C",
+                "spo2" => "%",
+                "urine_output_ml" => "ml",
+                _ => "",
+            };
+
+            metric_results.push(MetricTrendAnalysis {
+                metric: metric_name.to_string(),
+                unit: unit.to_string(),
+                data_point_count,
+                regression,
+                anomalies,
+            });
+        }
+
+        Ok(TrendAnalysisDto {
+            patient_id,
+            period_start: start_date.to_string(),
+            period_end: end_date.to_string(),
+            metrics: metric_results,
+        })
+    }
 }