feat(persona): RFM 八象限特征(统计层,additive)+ recompute --force

- 新增 rfm feature(融合 R最近/F频次/M金额三时间语义):data 带 segment 八象限 +
  lifecycle 生命周期 + valueTier(0-4)/riskScore(0-3)。统一旧 value+recall_risk。
- additive 接入(暂不动 scorer/旧特征);本地 928 验证:分布临床合理,且
  valueTier/riskScore 100% 复现旧 value.score/recall_risk.score → 后续翻转零风险。
- recompute-persona 加 --force:算法/特征变更后(数据没变)跳过水位幂等闸强制重算
  (部署到服务器也需要,否则全 noop)。

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

apps/pac-service/src/cli/recompute-persona.cli.ts

@@ -18,6 +18,7 @@ interface Args {
   host: string;
   pid?: string;
   pids?: string[]; // 多个 externalId(逗号分隔),定向重算受影响子集用
+  force?: boolean; // 跳过水位幂等闸:算法/特征变更后(数据没变)也强制重算
 }
 
 function parseArgs(argv: string[]): Args {
@@ -25,6 +26,7 @@ function parseArgs(argv: string[]): Args {
   for (const a of argv) {
     if (a.startsWith('--host=')) args.host = a.slice('--host='.length);
     else if (a.startsWith('--pid=')) args.pid = a.slice('--pid='.length);
+    else if (a === '--force') args.force = true;
     else if (a.startsWith('--pids=')) {
       args.pids = a.slice('--pids='.length).split(',').map((s) => s.trim()).filter(Boolean);
     }
@@ -65,6 +67,7 @@ async function bootstrap() {
         const r = await svc.recompute({
           patientId: p.id,
           source: `manual:cli`,
+          force: args.force,
         });
         logger.log(
           `  ${p.externalId.padEnd(6)} ${(p.name ?? '').padEnd(12)} v${r.version} features=${r.featureCount} status=${r.status} (${r.durationMs}ms)`,

apps/pac-service/src/modules/persona/features/feature.registry.ts

@@ -5,6 +5,7 @@ import { TreatmentChainStatusFeatureExtractor } from './treatment-chain-status.f
 import { RecallRiskFeatureExtractor } from './recall-risk.feature';
 import { DoNotContactStatusFeatureExtractor } from './do-not-contact-status.feature';
 import { EntitlementStatusFeatureExtractor } from './entitlement-status.feature';
+import { RfmFeatureExtractor } from './rfm.feature';
 
 /**
  * FeatureRegistry — 收集所有规则路径的 PersonaFeature 提取器。
@@ -23,7 +24,9 @@ export class FeatureRegistry {
     risk: RecallRiskFeatureExtractor,
     dnc: DoNotContactStatusFeatureExtractor,
     entitlement: EntitlementStatusFeatureExtractor,
+    rfm: RfmFeatureExtractor,
   ) {
-    this.extractors = [value, chain, risk, dnc, entitlement];
+    // rfm(统计层)additive 加入;下一步翻转 scorer 读 rfm.data 后,推翻 value/chain/risk
+    this.extractors = [value, chain, risk, dnc, entitlement, rfm];
   }
 }

apps/pac-service/src/modules/persona/features/rfm.feature.ts

+import { Injectable } from '@nestjs/common';
+import {
+  PersonaFeatureKey,
+  FactType,
+  FactKind,
+  lookupDxTreatment,
+} from '@pac/types';
+import type {
+  ActiveFact,
+  FeatureExtractor,
+  FeatureExtractorContext,
+  PersonaFeatureDraft,
+} from './feature.interface';
+
+/**
+ * rfm 价值分群(RFM 八象限 + 生命周期)— 统计层第一个特征
+ *
+ * ⭐ 设计依据 docs/algorithm/persona-design-v2.md:
+ *   - RFM 天然融合三种时间语义:R 最近(snapshot)/ F 频次(window-24m)/ M 金额(lifetime)。
+ *   - 统一旧 value(=M)+ recall_risk(=R+gap):一个特征产出 valueTier + riskScore 给打分,
+ *     再叠加 segment(八象限)+ lifecycle(生命周期)供圈人群/话术。data 带 segment → 天然 campaign-ready。
+ *   - 特征只产「事实」(data),分值映射是各消费场景的事(scorer 读 data 自己算 bonus)。
+ *
+ * 口径(默认值,上线后用扫描器看真实分布校准,集中在此):
+ *   R 最近   = 距最后一次临床/到诊事件天数(encounter/treatment(actual)/diagnosis/recommendation/visit_registration)
+ *   F 频次   = 近 24 月去重到诊「天数」(encounter/treatment(actual)/visit_registration)
+ *   M 金额   = 累计净消费 = Σpayment + Σrecharge − Σrefund(cents,全历史)
+ *   高/低阈  = R 高 ≤270d(还在召回节律内)· F 高 ≥4 次/2年 · M 高 ≥¥3,000(银卡+)
+ */
+@Injectable()
+export class RfmFeatureExtractor implements FeatureExtractor {
+  readonly key = PersonaFeatureKey.RFM;
+
+  // M 金额分档(与旧 value 一致,保证 valueBonus 行为不变)
+  private static readonly M_TIERS = [
+    { cents: 3000_000, tier: 4, mScore: 5 }, // ¥30k 钻
+    { cents: 1000_000, tier: 3, mScore: 5 }, // ¥10k 金
+    { cents: 300_000, tier: 2, mScore: 4 }, // ¥3k  银(M 高分界)
+    { cents: 50_000, tier: 1, mScore: 2 }, // ¥500 普通
+  ];
+
+  // 八象限:(R 高, F 高, M 高) → segment key + 中文
+  private static segmentOf(rHi: boolean, fHi: boolean, mHi: boolean) {
+    const m = mHi ? 'important' : 'general';
+    const v = m === 'important' ? '重要' : '一般';
+    if (rHi && fHi) return { key: `${m}_value`, zh: `${v}价值` };
+    if (!rHi && fHi) return { key: `${m}_retain`, zh: `${v}保持` };
+    if (rHi && !fHi) return { key: `${m}_develop`, zh: `${v}发展` };
+    return { key: `${m}_winback`, zh: `${v}挽留` };
+  }
+
+  private static dayBucket(days: number, edges: number[]): number {
+    // edges 升序;返回 5..1(越靠前越"好"/越近)
+    for (let i = 0; i < edges.length; i++) if (days <= edges[i]!) return 5 - i;
+    return 5 - edges.length;
+  }
+
+  extract(ctx: FeatureExtractorContext): PersonaFeatureDraft {
+    const get = (t: string) => ctx.factsByType.get(t) ?? [];
+    const now = ctx.now;
+    const DAY = 86400_000;
+
+    // ── M:累计净消费 ──
+    const moneyFactIds: string[] = [];
+    let monetaryCents = 0;
+    for (const f of get('payment_record')) {
+      monetaryCents += Number((f.content as Record<string, unknown>).amount_cents ?? 0);
+      moneyFactIds.push(f.id);
+    }
+    for (const f of get('recharge_record')) {
+      monetaryCents += Number((f.content as Record<string, unknown>).amount_cents ?? 0);
+      moneyFactIds.push(f.id);
+    }
+    for (const f of get('refund_record')) {
+      monetaryCents -= Number((f.content as Record<string, unknown>).amount_cents ?? 0);
+      moneyFactIds.push(f.id);
+    }
+    const mTier = RfmFeatureExtractor.M_TIERS.find((t) => monetaryCents >= t.cents);
+    const valueTier = mTier?.tier ?? 0; // 0-4,喂 valueBonus
+    const mScore = mTier?.mScore ?? 1; // 1-5
+    const mHigh = monetaryCents >= 300_000; // ¥3k 银卡+
+
+    // ── R / F:到诊事件 ──
+    const visitFacts: ActiveFact[] = [
+      ...get(FactType.ENCOUNTER_RECORD),
+      ...get(FactType.TREATMENT_RECORD).filter((f) => f.kind === FactKind.ACTUAL),
+      ...get(FactType.VISIT_REGISTRATION_RECORD),
+    ];
+    // R 最近还纳入诊断/建议(任意临床触点),F 只算物理到诊
+    const recencyFacts: ActiveFact[] = [
+      ...visitFacts,
+      ...get(FactType.DIAGNOSIS_RECORD),
+      ...get(FactType.RECOMMENDATION_RECORD),
+    ];
+    const clinicalFactIds = recencyFacts.map((f) => f.id);
+
+    let latest: Date | null = null;
+    let earliest: Date | null = null;
+    for (const f of recencyFacts) {
+      if (!f.occurredAt) continue;
+      if (!latest || f.occurredAt > latest) latest = f.occurredAt;
+      if (!earliest || f.occurredAt < earliest) earliest = f.occurredAt;
+    }
+    const recencyDays = latest ? Math.floor((now.getTime() - latest.getTime()) / DAY) : 9999;
+    const firstDays = earliest ? Math.floor((now.getTime() - earliest.getTime()) / DAY) : 9999;
+
+    // F:近 24 月去重到诊天数
+    const win = now.getTime() - 730 * DAY;
+    const visitDays = new Set<string>();
+    for (const f of visitFacts) {
+      if (!f.occurredAt || f.occurredAt.getTime() < win) continue;
+      visitDays.add(f.occurredAt.toISOString().slice(0, 10));
+    }
+    const freqCount24m = visitDays.size;
+
+    const rHigh = recencyDays <= 270;
+    const fHigh = freqCount24m >= 4;
+    const rScore = RfmFeatureExtractor.dayBucket(recencyDays, [90, 270, 540, 900]); // 5..1
+    const fScore = freqCount24m >= 8 ? 5 : freqCount24m >= 4 ? 4 : freqCount24m >= 2 ? 3 : freqCount24m >= 1 ? 2 : 1;
+
+    const seg = RfmFeatureExtractor.segmentOf(rHigh, fHigh, mHigh);
+
+    // ── lifecycle 生命周期(R 派生;reactivated 回流暂缓,后续加)──
+    let lifecycle: string;
+    if (recencyDays > 540) lifecycle = 'churned';
+    else if (recencyDays > 270) lifecycle = 'silent';
+    else if (firstDays <= 180) lifecycle = 'new';
+    else lifecycle = 'active';
+    const lifecycleZh = { churned: '流失', silent: '沉默', new: '新客', active: '活跃' }[lifecycle]!;
+
+    // ── riskScore 0-3(=旧 recall_risk:R + 治疗缺口),喂 likelihoodBonus ──
+    const actualCats = new Set<string>();
+    for (const tx of get(FactType.TREATMENT_RECORD)) {
+      if (tx.kind !== FactKind.ACTUAL) continue;
+      const cat = String((tx.content as Record<string, unknown>).category ?? '');
+      if (cat) actualCats.add(cat);
+    }
+    let hasGap = false;
+    for (const sig of [...get(FactType.DIAGNOSIS_RECORD), ...get(FactType.RECOMMENDATION_RECORD)]) {
+      if (!sig.occurredAt) continue;
+      const rule = lookupDxTreatment(String((sig.content as Record<string, unknown>).code ?? ''));
+      if (!rule || rule.categories.some((c) => actualCats.has(c))) continue;
+      if (Math.floor((now.getTime() - sig.occurredAt.getTime()) / DAY) > rule.windowDays) {
+        hasGap = true;
+        break;
+      }
+    }
+    let riskScore: number;
+    if (recencyDays >= 540 && hasGap) riskScore = 3;
+    else if (recencyDays >= 360 || hasGap) riskScore = 2;
+    else if (recencyDays >= 180) riskScore = 1;
+    else riskScore = 0;
+
+    const yuan = (monetaryCents / 100).toLocaleString('zh-CN', { maximumFractionDigits: 0 });
+    const recencyStr = latest ? `距上次 ${recencyDays} 天` : '无到诊记录';
+
+    return {
+      key: this.key,
+      // 人读自包含:分群 · 生命周期 · R/F/M 概要
+      description: `${seg.zh} · ${lifecycleZh} · ${recencyStr} · 近2年${freqCount24m}次 · 累计¥${yuan}`,
+      // score = 粗排提示(R+F+M 之和,1-5 各维),非语义载体;场景从 data 自算分
+      score: rScore + fScore + mScore,
+      data: {
+        segment: seg.key, // 八象限 key → 圈人群
+        lifecycle, // 生命周期阶段
+        rScore,
+        fScore,
+        mScore,
+        rHigh,
+        fHigh,
+        mHigh,
+        recencyDays,
+        firstSeenDays: firstDays,
+        freqCount24m,
+        monetaryCents,
+        valueTier, // 0-4 → 兼容 valueBonus
+        riskScore, // 0-3 → 兼容 likelihoodBonus
+        hasTreatmentGap: hasGap,
+      },
+      evidence: { factIds: [...new Set([...moneyFactIds, ...clinicalFactIds])] },
+    };
+  }
+}

apps/pac-service/src/modules/persona/persona.module.ts

@@ -8,6 +8,7 @@ import { TreatmentChainStatusFeatureExtractor } from './features/treatment-chain
 import { RecallRiskFeatureExtractor } from './features/recall-risk.feature';
 import { DoNotContactStatusFeatureExtractor } from './features/do-not-contact-status.feature';
 import { EntitlementStatusFeatureExtractor } from './features/entitlement-status.feature';
+import { RfmFeatureExtractor } from './features/rfm.feature';
 
 @Module({
   controllers: [PersonaController],
@@ -21,6 +22,7 @@ import { EntitlementStatusFeatureExtractor } from './features/entitlement-status
     RecallRiskFeatureExtractor,
     DoNotContactStatusFeatureExtractor,
     EntitlementStatusFeatureExtractor,
+    RfmFeatureExtractor,
   ],
   exports: [PersonaService],
 })

apps/pac-service/src/modules/persona/persona.service.ts

@@ -42,6 +42,7 @@ export class PersonaService {
     patientId: string;
     source: string; // 'txn:<id>' / 'manual:<userId>' / 'scheduled:<cronId>'
     now?: Date;
+    force?: boolean; // 跳过水位幂等闸:算法/特征变更后重算(数据没变也要重跑)
   }): Promise<RecomputeResult> {
     const startedAt = new Date();
     const now = input.now ?? new Date();
@@ -71,6 +72,7 @@ export class PersonaService {
     });
 
     if (
+      !input.force &&
       currentPersona &&
       latestTxnForGate &&
       currentPersona.eventWatermark !== null &&

packages/types/src/enums/index.ts

@@ -379,6 +379,9 @@ export const PersonaFeatureKey = {
   RECALL_RISK: 'recall_risk', // 流失/复发风险
   DO_NOT_CONTACT_STATUS: 'do_not_contact_status', // 不打扰状态(合规硬约束)
 
+  // 统计层(RFM 八象限 — 融合 R 最近/F 频次/M 金额三种时间语义;统一旧 value+recall_risk)
+  RFM: 'rfm', // 价值分群(RFM 八象限 + 生命周期;data 带 segment 供圈人群)
+
   // v1 候选(规则路径,业务方反馈后逐步上)
   ENTITLEMENT_STATUS: 'entitlement_status', // 权益身份(商保直付 / 医保 / 储值 / 私行;事实投影型,史+最近日期)
   INCOMPLETE_TREATMENT: 'incomplete_treatment', // 未完成治疗(治疗链有缺口)

packages/types/src/labels.ts

@@ -38,6 +38,7 @@ export const planScenarioLabel = (key: string): string =>
 // ─────────────────────────────────────────────────────────
 
 export const PERSONA_FEATURE_META: Record<string, { label: string; tone: Tone }> = {
+  [PersonaFeatureKey.RFM]: { label: '价值分群', tone: 'indigo' },
   [PersonaFeatureKey.VALUE]: { label: '患者价值', tone: 'indigo' },
   [PersonaFeatureKey.TREATMENT_CHAIN_STATUS]: { label: '治疗链状态', tone: 'amber' },
   [PersonaFeatureKey.RECALL_RISK]: { label: '流失风险', tone: 'emerald' },