Improve patient list OCR cleanup

2026-06-11 20:31:11 +08:00
parent fde8e11cd1
commit 939c34be19
3 changed files with 155 additions and 15 deletions
--- a/患者列表处理/数据处理工作区/02_患者列表OCR归档.py
+++ b/患者列表处理/数据处理工作区/02_患者列表OCR归档.py
@@ -463,12 +463,30 @@ def ocr_response_to_rows(response: dict[str, Any], engine: str, image_width: int

 def normalize_date(text: str) -> str:
    text = normalize_text(text).replace("/", "-").replace(".", "-")
-    text = re.sub(r"(\d{4})-(\d{1,2})-(\d{1,2})\s+(\d{1,2}):(\d{1,2}):(\d{1,2})", date_repl, text)
+    text = re.sub(
+        r"(\d{4})-(\d{1,2})-(\d{1,2})\s+(\d{1,4}):(\d{1,2})(?::(\d{1,2}))?",
+        date_repl,
+        text,
+    )
    return text


 def date_repl(match: re.Match[str]) -> str:
-    year, month, day, hour, minute, second = match.groups()
+    year, month, day, hour_text, minute_text, second_text = match.groups()
+    if second_text is None and len(hour_text) > 2:
+        if len(hour_text) == 3:
+            hour = int(hour_text[:2])
+            minute = int(hour_text[2:])
+        else:
+            hour = int(hour_text[:-2])
+            minute = int(hour_text[-2:])
+        second = int(minute_text)
+    else:
+        hour = int(hour_text)
+        minute = int(minute_text)
+        second = int(second_text or 0)
+    if not (0 <= hour <= 23 and 0 <= minute <= 59 and 0 <= second <= 59):
+        return match.group(0)
    return f"{int(year):04d}-{int(month):02d}-{int(day):02d} {int(hour):02d}:{int(minute):02d}:{int(second):02d}"


@@ -485,19 +503,24 @@ def clean_patient_row(row: list[str]) -> dict[str, Any]:
            values["性别"] = sex_age.group(1)
            if not values["年龄"]:
                values["年龄"] = sex_age.group(2)
-    values["住院号"] = re.sub(r"\s+", "", values["住院号"]).upper()
-    if values["住院号"].startswith("ZV"):
-        values["住院号"] = "ZY" + values["住院号"][2:]
-    if values["住院号"].startswith("ZYS"):
-        values["住院号"] = "ZY5" + values["住院号"][3:]
-    if len(values["住院号"]) > 2:
-        prefix, number = values["住院号"][:2], values["住院号"][2:]
-        values["住院号"] = prefix + number.translate(str.maketrans({"O": "0", "I": "1", "L": "1", "S": "5"}))
+    def normalize_inpatient_no(value: Any) -> str:
+        inpatient_no = re.sub(r"\s+", "", normalize_text(value)).upper()
+        if inpatient_no.startswith("ZV"):
+            inpatient_no = "ZY" + inpatient_no[2:]
+        if inpatient_no.startswith("ZYS"):
+            inpatient_no = "ZY5" + inpatient_no[3:]
+        if len(inpatient_no) > 2:
+            prefix, number = inpatient_no[:2], inpatient_no[2:]
+            inpatient_no = prefix + number.translate(str.maketrans({"O": "0", "I": "1", "L": "1", "S": "5"}))
+        return inpatient_no
+
+    values["住院号"] = normalize_inpatient_no(values["住院号"])
+
    def looks_datetime(value: Any) -> bool:
        return bool(
            re.fullmatch(
-                r"\d{4}[-/.]\d{1,2}[-/.]\d{1,2}\s+\d{1,2}:\d{1,2}:\d{1,2}",
-                str(value),
+                r"\d{4}-\d{2}-\d{2}\s+\d{2}:\d{2}:\d{2}",
+                normalize_date(str(value)),
            )
        )

@@ -510,8 +533,47 @@ def clean_patient_row(row: list[str]) -> dict[str, Any]:
    values["入院时间"] = normalize_date(values["入院时间"])
    values["最后书写时间"] = normalize_date(values["最后书写时间"])
    values["出院时间"] = normalize_date(values["出院时间"])
+
+    long_name_shift = re.fullmatch(r"(.+?)(?:\.\.\.)?\s*(男|女)", values["姓名"])
+    if (
+        long_name_shift
+        and values["性别"].endswith("岁")
+        and re.fullmatch(r"Z[YV]\w+", values["年龄"], re.IGNORECASE)
+        and values["住院号"]
+    ):
+        old_sex = values["性别"]
+        old_age = values["年龄"]
+        old_diagnosis = values["住院号"]
+        values["姓名"] = normalize_text(long_name_shift.group(1))
+        values["性别"] = long_name_shift.group(2)
+        values["年龄"] = old_sex
+        values["住院号"] = normalize_inpatient_no(old_age)
+        if looks_datetime(values["诊断"]) and looks_datetime(values["入院时间"]) and looks_days(values["最后书写时间"]):
+            old_admission = values["诊断"]
+            old_last_write = values["入院时间"]
+            old_days = values["最后书写时间"]
+            old_discharge = values["住院天数"]
+            old_postop = values["出院时间"]
+            values["诊断"] = old_diagnosis
+            values["入院时间"] = normalize_date(old_admission)
+            values["最后书写时间"] = normalize_date(old_last_write)
+            values["住院天数"] = old_days
+            values["出院时间"] = normalize_date(old_discharge) if looks_datetime(old_discharge) else ""
+            if looks_postop(old_postop) and not values["手术后天数"]:
+                values["手术后天数"] = old_postop
+        elif not values["诊断"]:
+            values["诊断"] = old_diagnosis
+
+    admission_with_diagnosis = re.search(
+        r"(.+?)\s*-*\s*(\d{4}[-/.]\d{1,2}[-/.]\d{1,2}\s+\d{1,4}:\d{1,2}(?::\d{1,2})?)\s*$",
+        values["入院时间"],
+    )
+    if admission_with_diagnosis and not looks_datetime(values["入院时间"]):
+        values["诊断"] = normalize_text(f"{values['诊断']} {admission_with_diagnosis.group(1)}")
+        values["入院时间"] = normalize_date(admission_with_diagnosis.group(2))
+
    shifted_by_long_diagnosis = re.search(
-        r"(\d{4}[-/.]\d{1,2}[-/.]\d{1,2}\s+\d{1,2}:\d{1,2}:\d{1,2})$",
+        r"(\d{4}[-/.]\d{1,2}[-/.]\d{1,2}\s+\d{1,4}:\d{1,2}(?::\d{1,2})?)\s*[)）\]]*$",
        values["诊断"],
    )
    if (
@@ -539,6 +601,25 @@ def clean_patient_row(row: list[str]) -> dict[str, Any]:
        values["诊断"] = values["诊断"][: shifted_by_long_diagnosis.start()].strip()
        values["入院时间"] = normalize_date(shifted_by_long_diagnosis.group(1))

+    if (
+        not looks_datetime(values["入院时间"])
+        and looks_datetime(values["最后书写时间"])
+        and looks_datetime(values["住院天数"])
+        and looks_days(values["出院时间"])
+    ):
+        diagnosis_extra = values["入院时间"]
+        old_admission = values["最后书写时间"]
+        old_last_write = values["住院天数"]
+        old_days = values["出院时间"]
+        old_discharge = values["手术后天数"]
+        if diagnosis_extra:
+            values["诊断"] = normalize_text(f"{values['诊断']} {diagnosis_extra}")
+        values["入院时间"] = normalize_date(old_admission)
+        values["最后书写时间"] = normalize_date(old_last_write)
+        values["住院天数"] = old_days
+        values["出院时间"] = normalize_date(old_discharge) if looks_datetime(old_discharge) else ""
+        values["手术后天数"] = ""
+
    days_with_discharge = re.fullmatch(
        r"(\d+)\s+(\d{4}[-/.]\d{1,2}[-/.]\d{1,2}\s+\d{1,2}:\d{1,2}:\d{1,2})",
        str(values["住院天数"]),
@@ -551,6 +632,10 @@ def clean_patient_row(row: list[str]) -> dict[str, Any]:
    if looks_postop(values["出院时间"]) and not values["手术后天数"]:
        values["手术后天数"] = values["出院时间"]
        values["出院时间"] = ""
+    if values["出院时间"] in {"手术日", "当日", "当天"}:
+        if not values["手术后天数"]:
+            values["手术后天数"] = values["出院时间"]
+        values["出院时间"] = ""
    if looks_postop(values["住院天数"]) and not values["手术后天数"]:
        values["手术后天数"] = values["住院天数"]
        values["住院天数"] = ""
@@ -694,6 +779,7 @@ def apply_corrections(records: list[dict[str, Any]], corrections: dict[tuple[str
            continue
        correction = corrections[key]
        record["患者信息"].update(correction.get("患者信息", {}))
+        record["图片信息"].update(correction.get("图片信息", {}))
        record["复核"]["人工修正"] = True
        record["复核"]["复核选项"] = correction.get("复核选项", {})
        if correction.get("复核备注"):
--- a/患者列表处理/数据处理工作区/05_同步PostgreSQL单表.py
+++ b/患者列表处理/数据处理工作区/05_同步PostgreSQL单表.py
@@ -12,6 +12,7 @@ import argparse
 import csv
 import json
 import os
+import re
 import subprocess
 import tempfile
 from pathlib import Path
@@ -22,6 +23,7 @@ DEFAULT_RESULT_JSON = Path("数据处理结果区/合并_患者列表_结构化.
 DEFAULT_SCHEMA_SQL = Path("数据处理工作区/06_PostgreSQL建表结构.sql")

 TABLE_NAME = '"Patient_Lists"'
+DATE_TIME_RE = re.compile(r"^\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}$")

 PATIENT_FIELDS = [
    "batch_name",
@@ -75,6 +77,16 @@ def review_note_items(review: dict[str, Any]) -> list[str]:
    return note_items


+def admission_after_discharge(admission_time: Any, discharge_time: Any) -> bool:
+    admission = scalar(admission_time).strip()
+    discharge = scalar(discharge_time).strip()
+    if not admission or not discharge:
+        return False
+    if not DATE_TIME_RE.match(admission) or not DATE_TIME_RE.match(discharge):
+        return False
+    return admission > discharge
+
+
 def psql_quote(path: Path) -> str:
    return "'" + str(path.resolve()).replace("'", "''") + "'"

@@ -87,6 +99,17 @@ def flatten_patient_record(record: dict[str, Any]) -> dict[str, Any]:
    hospital_days, hospital_days_note = clean_integer(patient["住院天数"])
    if hospital_days_note:
        note_items.append(hospital_days_note)
+    review_status = review["状态"]
+    patient_name = scalar(patient["姓名"]).strip()
+    if not patient_name:
+        patient_name = "缺少姓名"
+        review_status = "需人工复核"
+        if "缺少姓名" not in note_items:
+            note_items.append("缺少姓名")
+    if admission_after_discharge(patient["入院时间"], patient["出院时间"]):
+        review_status = "需人工复核"
+        if "入院时间晚于出院时间" not in note_items:
+            note_items.append("入院时间晚于出院时间")
    review_notes = "；".join(note_items)
    return {
        "batch_name": record["处理批次"],
@@ -96,7 +119,7 @@ def flatten_patient_record(record: dict[str, Any]) -> dict[str, Any]:
        "image_path": image["图片路径"],
        "image_name": image["图片名"],
        "image_row_no": image["图片内行号"],
-        "patient_name": patient["姓名"],
+        "patient_name": patient_name,
        "gender": patient["性别"],
        "age": patient["年龄"],
        "inpatient_no": patient["住院号"],
@@ -106,7 +129,7 @@ def flatten_patient_record(record: dict[str, Any]) -> dict[str, Any]:
        "hospital_days": hospital_days,
        "discharge_time": patient["出院时间"],
        "postoperative_days": patient["手术后天数"],
-        "review_status": review["状态"],
+        "review_status": review_status,
        "review_notes": review_notes,
        "manual_corrected": bool(review.get("人工修正")),
    }
--- a/患者列表处理/数据处理工作区/06_PostgreSQL建表结构.sql
+++ b/患者列表处理/数据处理工作区/06_PostgreSQL建表结构.sql
@@ -49,6 +49,37 @@ CREATE INDEX IF NOT EXISTS idx_patient_lists_patient_name ON "Patient_Lists"(pat
 CREATE INDEX IF NOT EXISTS idx_patient_lists_review_status ON "Patient_Lists"(review_status);
 CREATE INDEX IF NOT EXISTS idx_patient_lists_audit_result ON "Patient_Lists"(audit_result);

+DO $$
+BEGIN
+  IF NOT EXISTS (
+    SELECT 1 FROM pg_constraint
+    WHERE conrelid = '"Patient_Lists"'::regclass
+      AND conname = 'ck_patient_lists_inpatient_no_present'
+  ) THEN
+    ALTER TABLE "Patient_Lists"
+      ADD CONSTRAINT ck_patient_lists_inpatient_no_present
+      CHECK (btrim(inpatient_no) <> '') NOT VALID;
+  END IF;
+
+  IF NOT EXISTS (
+    SELECT 1 FROM pg_constraint
+    WHERE conrelid = '"Patient_Lists"'::regclass
+      AND conname = 'ck_patient_lists_admission_before_discharge'
+  ) THEN
+    ALTER TABLE "Patient_Lists"
+      ADD CONSTRAINT ck_patient_lists_admission_before_discharge
+      CHECK (
+        COALESCE(discharge_time, '') = ''
+        OR COALESCE(admission_time, '') = ''
+        OR admission_time !~ '^\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}$'
+        OR discharge_time !~ '^\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}$'
+        OR admission_time <= discharge_time
+        OR review_status = '需人工复核'
+      ) NOT VALID;
+  END IF;
+END
+$$;
+
 COMMENT ON TABLE "Patient_Lists" IS 'HIS患者列表图片OCR归档后的正式患者记录单表';
 COMMENT ON COLUMN "Patient_Lists".record_id IS '患者记录主键，数据库自动生成';
 COMMENT ON COLUMN "Patient_Lists".batch_name IS '处理批次名称，通常为原始图片集群文件夹名';