File size: 10,143 Bytes
089d665 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 | """DATASUS SIM (mortality) pull pipeline.
For training NeuralSurv on REAL Brazilian rare-disease mortality data.
Pulls DOXX####.dbc for given UFs and years from
ftp://ftp.datasus.gov.br/dissemin/publicos/SIM/CID10/DORES/, parses with
pyreaddbc, filters to rare-disease CIDs, extracts (sex, age, UF,
cause_cid, date_of_death, date_of_birth) tuples.
Output: pandas DataFrame ready for survival analysis.
"""
from __future__ import annotations
import logging
import os
import tempfile
import urllib.request
from datetime import datetime
from pathlib import Path
logger = logging.getLogger("gemeo.datasus.sim")
# Rare disease CID-10 codes (CID-10-BR encoding strips the dot in DBC)
# Maps DBC code → ORPHA code for our seeded diseases
RARE_CIDS_CID10 = {
"G113": "100", # AT (G11.3) — though sometimes G11.3 is generic ataxia
"E752": ["646", "355"], # NPC + Gaucher (E75.2 sphingolipidosis)
"E751": ["355"], # Gaucher specifically
"G710": "98896", # DMD (G71.0)
"G120": "70", # SMA-1 (G12.0)
"E840": "586", "E841": "586", "E848": "586", "E849": "586", # CF
"E760": "579", # MPS I (E76.0)
"E761": "580", # MPS II (E76.1)
"E83.0": "905", "E830": "905", # Wilson
"G11.1": "95", "G111": "95", # Friedreich
"Q874": "558", # Marfan (Q87.4)
"Q850": "636", # NF1 (Q85.0)
"F842": "778", # Rett (F84.2)
"D811": "183660", # SCID (D81.1)
}
# Brazilian UFs
ALL_UFS = ["AC", "AL", "AP", "AM", "BA", "CE", "DF", "ES", "GO", "MA",
"MT", "MS", "MG", "PA", "PB", "PR", "PE", "PI", "RJ", "RN",
"RS", "RO", "RR", "SC", "SP", "SE", "TO"]
def parse_age_idade(idade_str: str) -> float:
"""Parse SIM IDADE field. Format: prefix + value.
0XX = minutes/seconds (newborn)
1XX = hours
2XX = days
3XX = months
4XX = years
5XX = 100+ years (XX is years - 100)
"""
if not idade_str or len(idade_str) < 2:
return None
try:
prefix = int(idade_str[0])
val = int(idade_str[1:].lstrip("0") or "0")
except (ValueError, IndexError):
return None
if prefix == 0: # minutes
return val / 525600
if prefix == 1: # hours
return val / 8760
if prefix == 2: # days
return val / 365.25
if prefix == 3: # months
return val / 12
if prefix == 4: # years
return float(val)
if prefix == 5: # 100+ years
return 100.0 + val
return None
def parse_date_yyyymmdd(date_str: str):
"""Parse SIM date field (DDMMYYYY format)."""
if not date_str or len(date_str) != 8:
return None
try:
return datetime.strptime(date_str, "%d%m%Y").date()
except (ValueError, TypeError):
return None
def parse_sim_record(rec: dict) -> dict | None:
"""Parse a single SIM record into a clean dict."""
cid = (rec.get("CAUSABAS") or "").strip().upper()
if not cid:
return None
# Map to rare-disease ORPHA(s)
matched_orpha = RARE_CIDS_CID10.get(cid) or RARE_CIDS_CID10.get(cid + "0")
if matched_orpha is None:
# Try with dot
for k in (cid[:3] + "." + cid[3:], cid):
if k in RARE_CIDS_CID10:
matched_orpha = RARE_CIDS_CID10[k]
break
if matched_orpha is None:
return None
if isinstance(matched_orpha, list):
matched_orpha = matched_orpha[0]
age_yr = parse_age_idade((rec.get("IDADE") or "").strip())
sex_code = str(rec.get("SEXO") or "").strip()
sex = "M" if sex_code == "1" else ("F" if sex_code == "2" else "?")
uf_code = (rec.get("CODMUNRES") or "").strip()[:2]
return {
"cid": cid,
"orpha": matched_orpha,
"age_at_death_years": age_yr,
"sex": sex,
"uf_code": uf_code,
"date_of_death": parse_date_yyyymmdd((rec.get("DTOBITO") or "").strip()),
"date_of_birth": parse_date_yyyymmdd((rec.get("DTNASC") or "").strip()),
"race": rec.get("RACACOR"),
"education": rec.get("ESC"),
}
def pull_sim(uf: str, year: int, *, cache_dir: str = None,
target_cids: set = None) -> list[dict]:
"""Pull SIM for one UF/year, return parsed records matching target CIDs.
Args:
uf: 2-letter UF code (SP, RJ, MG, etc.)
year: 4-digit year
cache_dir: optional persistent cache; defaults to tempdir
target_cids: set of CID-10 codes (without dot) to filter; if None
uses RARE_CIDS_CID10
Returns:
list of parsed record dicts (each = parse_sim_record output)
"""
import pyreaddbc
from dbfread import DBF
if target_cids is None:
target_cids = set(RARE_CIDS_CID10.keys())
fname = f"DO{uf}{year}.dbc"
url = f"ftp://ftp.datasus.gov.br/dissemin/publicos/SIM/CID10/DORES/{fname}"
use_persistent = cache_dir is not None
if use_persistent:
os.makedirs(cache_dir, exist_ok=True)
dbc_path = os.path.join(cache_dir, fname)
dbf_path = dbc_path.replace(".dbc", ".dbf")
# Skip download if already cached and non-empty
if os.path.exists(dbf_path) and os.path.getsize(dbf_path) > 1024:
logger.info(f" [{uf}/{year}] cached: {fname}")
elif os.path.exists(dbc_path) and os.path.getsize(dbc_path) > 1024:
pyreaddbc.dbc2dbf(dbc_path, dbf_path)
else:
logger.info(f" [{uf}/{year}] download {url}")
urllib.request.urlretrieve(url, dbc_path)
pyreaddbc.dbc2dbf(dbc_path, dbf_path)
else:
td = tempfile.mkdtemp()
dbc_path = os.path.join(td, fname)
dbf_path = dbc_path.replace(".dbc", ".dbf")
try:
urllib.request.urlretrieve(url, dbc_path)
pyreaddbc.dbc2dbf(dbc_path, dbf_path)
except Exception as e:
logger.warning(f" [{uf}/{year}] download/convert failed: {e}")
return []
out = []
try:
for rec in DBF(dbf_path, encoding="latin-1", load=False):
cid_short = (rec.get("CAUSABAS") or "").strip().upper()
if cid_short not in target_cids:
continue
parsed = parse_sim_record(rec)
if parsed:
parsed["year"] = year
out.append(parsed)
except Exception as e:
logger.warning(f" [{uf}/{year}] parse failed: {e}")
logger.info(f" [{uf}/{year}] matched {len(out)} records")
return out
def pull_sim_multi(ufs: list[str], years: list[int], *,
cache_dir: str = None) -> list[dict]:
"""Pull SIM across multiple UFs and years, aggregate."""
all_records = []
for year in years:
for uf in ufs:
try:
recs = pull_sim(uf, year, cache_dir=cache_dir)
all_records.extend(recs)
except Exception as e:
logger.warning(f" [{uf}/{year}] error: {e}")
return all_records
def survival_distributions(records: list[dict]) -> dict:
"""Compute per-disease survival statistics from SIM records."""
from collections import defaultdict
import statistics
by_orpha = defaultdict(list)
for r in records:
if r.get("age_at_death_years") is not None:
by_orpha[r["orpha"]].append(r["age_at_death_years"])
out = {}
for orpha, ages in by_orpha.items():
if len(ages) < 3:
out[orpha] = {"n": len(ages), "ages": ages, "median": None}
continue
ages_sorted = sorted(ages)
n = len(ages_sorted)
median = statistics.median(ages_sorted)
p25 = ages_sorted[n // 4]
p75 = ages_sorted[3 * n // 4]
iqr = p75 - p25
# Weibull fit (lightweight method-of-moments)
try:
mean = sum(ages_sorted) / n
var = sum((a - mean) ** 2 for a in ages_sorted) / n
cv = (var ** 0.5) / mean if mean > 0 else 1.0
shape = 1.2 / cv if cv > 0 else 1.5
scale = mean / 0.91 # rough approximation
except Exception:
shape = 1.5
scale = median * 1.4
out[orpha] = {
"n": n,
"median": round(median, 2),
"p25": round(p25, 2),
"p75": round(p75, 2),
"iqr": round(iqr, 2),
"min_age": round(min(ages_sorted), 2),
"max_age": round(max(ages_sorted), 2),
"weibull_shape": round(shape, 2),
"weibull_scale": round(scale, 2),
"ages": ages_sorted,
}
return out
if __name__ == "__main__":
import argparse
import json
logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(name)s] %(message)s")
parser = argparse.ArgumentParser()
parser.add_argument("--ufs", nargs="+", default=["SP"])
parser.add_argument("--years", nargs="+", type=int, default=[2018, 2019, 2020])
parser.add_argument("--cache-dir", default="/tmp/datasus_cache")
parser.add_argument("--out-json", default="/tmp/datasus_survival.json")
args = parser.parse_args()
print(f"Pulling SIM for UFs={args.ufs} years={args.years}")
recs = pull_sim_multi(args.ufs, args.years, cache_dir=args.cache_dir)
print(f"\nTotal rare-CID records: {len(recs)}")
survival = survival_distributions(recs)
print(f"\nPer-disease survival distributions ({len(survival)} diseases):\n")
for orpha, s in sorted(survival.items()):
if s.get("median"):
print(f" ORPHA:{orpha:>6} n={s['n']:>4} median={s['median']:>6}y "
f"IQR=[{s['p25']:.0f}-{s['p75']:.0f}] Weibull(shape={s['weibull_shape']}, scale={s['weibull_scale']})")
else:
print(f" ORPHA:{orpha:>6} n={s['n']:>4} (insufficient for fit)")
# Save
out = {"records_count": len(recs), "ufs": args.ufs, "years": args.years,
"survival": {k: {kk: vv for kk, vv in v.items() if kk != "ages"}
for k, v in survival.items()},
"raw_sample": recs[:50]}
with open(args.out_json, "w") as f:
json.dump(out, f, default=str, indent=2)
print(f"\nSaved → {args.out_json}")
|