Informe de Búsqueda Bibliográfica — NeuroTrack AI

"""
NeuroTrack AI — PubMed Search via NCBI E-utilities
Búsqueda reproducible de literatura biomédica sobre IA + Alzheimer.
Requiere: requests, python-dotenv
Variables de entorno: NCBI_EMAIL, NCBI_API_KEY
"""

import os
import time
import json
import csv
from datetime import date
import requests

BASE = "https://eutils.ncbi.nlm.nih.gov/entrez/eutils"
TODAY = date.today().isoformat()

# ─── Configuración ─────────────────────────────────────────────────────────
NCBI_EMAIL   = os.environ.get("NCBI_EMAIL", "investigacion@neurotrack.ai")
NCBI_API_KEY = os.environ.get("NCBI_API_KEY")  # obtener en: ncbi.nlm.nih.gov/account/
RETMAX       = 200

# ─── Queries de búsqueda (3 passes) ────────────────────────────────────────
SEARCH_PASSES = [
    {
        "id": "pass1",
        "label": "IA + Alzheimer (revisiones sistemáticas)",
        "query": (
            "(artificial intelligence[mh] OR machine learning[tiab] OR deep learning[tiab])"
            " AND Alzheimer disease[mh]"
            " AND diagnosis[sh]"
            " AND 2022:2026[dp]"
            " AND english[la]"
            " AND (systematic review[pt] OR meta-analysis[pt])"
        ),
    },
    {
        "id": "pass2",
        "label": "Biomarcadores plasmáticos + ML",
        "query": (
            "(amyloid beta-peptides[mh] OR tau proteins[mh])"
            " AND (machine learning[tiab] OR neural network[tiab] OR random forest[tiab])"
            " AND blood[tiab]"
            " AND 2022:2026[dp]"
            " AND english[la]"
            " AND hasabstract[text]"
        ),
    },
    {
        "id": "pass3",
        "label": "MRI + CNN/Transformer + Sensibilidad/Especificidad",
        "query": (
            "magnetic resonance imaging[mh]"
            " AND (convolutional neural network[tiab] OR CNN[tiab] OR transformer[tiab])"
            " AND Alzheimer disease[mh]"
            " AND sensitivity AND specificity[mh]"
            " AND 2022:2026[dp]"
            " AND free full text[sb]"
        ),
    },
]

# ─── Funciones E-utilities ──────────────────────────────────────────────────
def _params(**extra) -> dict:
    p = {"tool": "neurotrack-pubmed", "email": NCBI_EMAIL}
    if NCBI_API_KEY:
        p["api_key"] = NCBI_API_KEY
    p.update(extra)
    return p


def esearch(query: str, retmax: int = RETMAX) -> tuple[list[str], int]:
    r = requests.get(
        f"{BASE}/esearch.fcgi",
        params=_params(db="pubmed", term=query,
                         retmode="json", retmax=retmax),
        timeout=30,
    )
    r.raise_for_status()
    time.sleep(0.35)
    data = r.json()["esearchresult"]
    return data["idlist"], int(data["count"])


def efetch_abstracts(pmids: list[str]) -> str:
    r = requests.get(
        f"{BASE}/efetch.fcgi",
        params=_params(db="pubmed", id=",".join(pmids),
                         rettype="abstract", retmode="text"),
        timeout=60,
    )
    r.raise_for_status()
    time.sleep(0.35)
    return r.text


# ─── Ejecución principal ────────────────────────────────────────────────────
log_rows = []

for pass_ in SEARCH_PASSES:
    print(f"[{pass_['id']}] Ejecutando: {pass_['label']}")
    pmids, total = esearch(pass_["query"])
    print(f"  → {total} resultados totales | {len(pmids)} recuperados")

    log_rows.append({
        "pass_id":       pass_["id"],
        "label":         pass_["label"],
        "database":      "PubMed/MEDLINE",
        "date_searched": TODAY,
        "query":         pass_["query"],
        "total_results": total,
        "retrieved":     len(pmids),
        "pmids":         "|".join(pmids),
    })

# Guardar log CSV
with open("pubmed_search_log.csv", "w", newline="", encoding="utf-8") as f:
    writer = csv.DictWriter(f, fieldnames=log_rows[0].keys())
    writer.writeheader()
    writer.writerows(log_rows)

print("\n✓ Log guardado en pubmed_search_log.csv")