code/data_processing/data_preV1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "883f7665",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['id', 'original_text_language', 'source_topic', 'readability_versions'])"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import json\n",
    "path=\"/home/mshahidul/readctrl/dataset_buildup.json\"\n",
    "lang=path.split(\"/\")[-1].split(\"_\")[0]\n",
    "with open(f\"{path}\", \"r\") as f:\n",
    "    data = json.load(f)\n",
    "\n",
    "data[0].keys()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "964139ea",
   "metadata": {},
   "source": [
    "## fernandez_huerta score calculation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "25449b20",
   "metadata": {},
   "outputs": [],
   "source": [
    "from FH_es import fernandez_huerta\n",
    "from FH_fr import flesch_kandel_moles_fr\n",
    "# from FH_pt import flesch_portuguese\n",
    "full_data=[]\n",
    "for item in data:\n",
    "    text = item[\"synthetic_summary\"]\n",
    "    dat={}\n",
    "    fh_score_b1 = fernandez_huerta(text['B1'])\n",
    "    fh_score_b2 = fernandez_huerta(text['B2'])\n",
    "    fh_score_b3 = fernandez_huerta(text['B3'])\n",
    "    dat['B1']={\n",
    "        \"text\": text['B1'],\n",
    "        \"fh_score\": fh_score_b1\n",
    "    }\n",
    "    dat['B2']={\n",
    "        \"text\": text['B2'],\n",
    "        \"fh_score\": fh_score_b2\n",
    "    }\n",
    "    dat['B3']={\n",
    "        \"text\": text['B3'],\n",
    "        \"fh_score\": fh_score_b3\n",
    "    }\n",
    "    full_data.append({\n",
    "        \"article\": item[\"article\"],\n",
    "        \"gold_summary\": item[\"gold_summary\"],\n",
    "        \"synthetic_summary\": dat\n",
    "    })\n",
    "with open(\"/home/mshahidul/readctrl/generating_data/score/synthetic.json\", \"w\") as f:\n",
    "    json.dump(full_data, f,indent=4, ensure_ascii=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "d61f82e4",
   "metadata": {},
   "outputs": [],
   "source": [
    "import textstat\n",
    "from FH_esV2 import fernandez_huerta\n",
    "# from FH_es import fernandez_huerta\n",
    "from FH_fr import flesch_kandel_moles_fr\n",
    "# from FH_pt import flesch_portuguese\n",
    "full_data=[]\n",
    "for item in data:\n",
    "    text = item[\"readability_versions\"]\n",
    "    dat={}\n",
    "    fh_score_easy = fernandez_huerta(text['easy']['text'])\n",
    "    fh_score_intermediate = fernandez_huerta(text['intermediate']['text'])\n",
    "    fh_score_hard = fernandez_huerta(text['hard']['text'])\n",
    "    dat['easy']={\n",
    "        \"text\": text['easy']['text'],\n",
    "        \"fh_score\": fh_score_easy\n",
    "    }\n",
    "    dat['intermediate']={\n",
    "        \"text\": text['intermediate']['text'],\n",
    "        \"fh_score\": fh_score_intermediate\n",
    "    }\n",
    "    dat['hard']={\n",
    "        \"text\": text['hard']['text'],\n",
    "        \"fh_score\": fh_score_hard\n",
    "    }\n",
    "    full_data.append({\n",
    "        \"original_text_language\": item[\"original_text_language\"],\n",
    "        \"source_topic\": item[\"source_topic\"],\n",
    "        \"synthetic_summary\": dat\n",
    "    })\n",
    "with open(\"/home/mshahidul/readctrl/generating_data/score/synthetic.json\", \"w\") as f:\n",
    "    json.dump(full_data, f,indent=4, ensure_ascii=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "23830e3d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 700x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Flatten fh_scores into a tidy DataFrame\n",
    "rows = []\n",
    "for item in full_data:\n",
    "    syn = item.get('synthetic_summary', {})\n",
    "    for level in ('easy', 'intermediate', 'hard'):\n",
    "        fh = syn.get(level, {}).get('fh_score')\n",
    "        if fh is not None:\n",
    "            rows.append({'level': level, 'fh_score': fh})\n",
    "\n",
    "df = pd.DataFrame(rows).dropna(subset=['fh_score'])\n",
    "\n",
    "# Plot\n",
    "sns.set_theme(style='whitegrid')\n",
    "plt.figure(figsize=(7,5))\n",
    "ax = sns.boxplot(data=df, x='level', y='fh_score')\n",
    "sns.stripplot(data=df, x='level', y='fh_score', color='black', alpha=0.4, jitter=0.2, size=3)\n",
    "ax.set_title(f'{lang} scores by level')\n",
    "ax.set_xlabel('Level')\n",
    "ax.set_ylabel('score')\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "16163d1e",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"/home/mshahidul/readctrl/generating_data/tik_ache/es_syntheticV3.json\", \"r\", encoding=\"utf-8\") as f:\n",
    "    tik_ache_data = json.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "8447f0ec",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "id: 40\n",
      "full text:\n",
      "\n",
      "La paciente era una recién nacida de 0 días de edad, de etnia Han, que nació a las 36 semanas de gestación de una mujer embarazada de 3, para 2, con un peso de 2650 g. Se le practicó una cesárea de urgencia debido a la angustia fetal, con puntuaciones de Apgar de 6 a 1 min y 5 min. Su grupo sanguíneo era O y RhD positivo. Al nacer, presentaba palidez, equimosis dispersas en múltiples áreas del cuerpo, sangrado de las mucosas e insuficiencia respiratoria, con un leve flujo de sangre hemorrágica también observado en los sitios de punción venosa. El análisis de gases en sangre de la arteria umbilical mostró un hematocrito de 0.08 y una hemoglobina de 23 g/L. La paciente requirió intubación endotraqueal y ventilación mecánica. Después de la transfusión de una suspensión de glóbulos rojos, los recuentos sanguíneos iniciales revelaron una trombocitopenia severa (recuento de plaquetas de 12 × 109/L), anemia (hemoglobina de 46 g/L) y leucopenia (recuento de leucocitos de 1.11 × 109/L).\n",
      "\n",
      "La paciente fue ingresada en la unidad de cuidados intensivos neonatales a las 3 horas de vida para recibir tratamiento adicional. Durante la hospitalización, se le diagnosticó coagulación intravascular diseminada (CID), con tiempo de tromboplastina parcial activada (TTPA) de 73,10 segundos, tiempo de protrombina (TP) de 25,4 segundos, fibrinógeno (FIB) de 1,01 g/L, razón internacional normalizada (INR) de 2,26 y dímero D > 20 mg/L. Se le administró ventilación mecánica, corrección de la acidosis y transfusión de plasma fresco congelado. A pesar de múltiples transfusiones de glóbulos rojos y plaquetas, los niveles de hemoglobina y plaquetas permanecieron por debajo de lo normal (hemoglobina 106 g/L y plaquetas 11 × 109/L el día 3). También recibió tratamiento antiinfeccioso, cardiotónico, vasopresor y otro tratamiento sintomático. Un frotis periférico mostró áreas de tinción pálida agrandadas de glóbulos rojos, con una proporción de reticulocitos del 1,5% y un resultado negativo en la prueba directa de Coombs. No se realizó aspiración de médula ósea debido a su grave estado. No hubo evidencia clínica o de laboratorio de sepsis neonatal, y las pruebas para toxoplasma, rubéola, citomegalovirus y virus del herpes simple (TORCH); virus de hepatitis; y Treponema pallidum fueron negativas. El examen físico de admisión reveló un estado mental deficiente, disminución del tono muscular en las extremidades y reflejo pupilar lento a la luz. Todos los demás hallazgos fueron normales. El ultrasonido craneal y electroencefalograma de cabecera revelaron hemorragia intracraneal grave y bajo voltaje, respectivamente. Los hallazgos del ecocardiograma sugirieron un conducto arterioso patente, foramen oval permeable e hipertensión pulmonar. El ultrasonido abdominal indicó hemorragia gastrointestinal. A pesar de todos los esfuerzos, la paciente falleció el tercer día de vida debido a falla multiorgánica y hemorragia intracraneal masiva (la información detallada se puede encontrar en el Informe Suplementario).\n",
      "\n",
      "Los padres de la paciente no eran consanguíneos y ambos tenían talasemia. La madre, que compartía el mismo tipo de sangre que la paciente, tuvo una anemia leve durante el embarazo (nivel de hemoglobina de 97 g/L) y un historial de aborto inducido. Los exámenes prenatales no mostraron anomalías, ni tampoco hidrops fetal, y no recibió ninguna medicación durante el embarazo que pudiera provocar una enfermedad hemorrágica de inicio temprano en el recién nacido. La paciente también tenía un hermano de 1 año que gozaba de buena salud. Su abuelo tenía un historial de anemia leve (detalles desconocidos).\n",
      "\n",
      "Se obtuvo el consentimiento informado de los padres del paciente y se realizó un secuenciamiento de Sanger para descubrir la causa de la enfermedad. Se detectó una nueva mutación de MECOM de desplazamiento de marco heterocigótica [NM_001105078: c.157_158del (p.Met53Glyfs*2)] en el probando. Esta variante cambió el aminoácido 53 de metionina (codón ATG) a glicina (codón GGT), seguido de una terminación temprana. La mutación no se encontró en los padres o en el hermano mayor. La variante se clasificó como patogénica según las directrices del Colegio Americano de Genética Médica (ACMG) [14]. El algoritmo “AutoPVS1” brindó un fuerte apoyo para la interpretación de PVS1 de p.Met53Glyfs*2, lo que indica patogenicidad. La variante no se ha informado previamente en la Base de Datos de Mutación Genética Humana (HGMD) o en Clinvar. El análisis de conservación mostró que el residuo Met53 está altamente conservado en todas las especies de mamíferos (incluidos humanos, ratones, ratas, chimpancés y bovinos) utilizando Clustal Omega. Se generaron modelos de estructura de proteínas tridimensionales de las proteínas MECOM de tipo salvaje y mutantes utilizando SWISS-MODEL, lo que indica que la mutación de desplazamiento de marco causó una terminación temprana de la síntesis de aminoácidos, alterando significativamente la estructura de la proteína.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "id=40\n",
    "print(f\"id: {id}\")\n",
    "print(\"full text:\\n\")\n",
    "print(tik_ache_data[id]['article'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "d802cf9e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['id', 'original_text_language', 'source_topic', 'readability_versions'])\n"
     ]
    }
   ],
   "source": [
    "with open(\"/home/mshahidul/readctrl/dataset_buildup.json\", \"r\", encoding=\"utf-8\") as f:\n",
    "    dataset_buildup = json.load(f)\n",
    "print(dataset_buildup[0].keys())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "05569bfb",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_prompts1={\n",
    "\"easy\":'''\n",
    "Reescribe el siguiente informe médico en español con un lenguaje sencillo y claro. Usa oraciones cortas, evita tecnicismos y explica los términos médicos con palabras comunes para que una persona sin formación médica lo comprenda fácilmente. Mantén los datos médicos esenciales.\n",
    "''',\n",
    "\"intermediate\": '''\n",
    "Reformula el siguiente informe médico en español en un nivel de lectura intermedio. Usa un lenguaje comprensible para personas con cultura general y cierto conocimiento de temas de salud. Mantén la precisión médica, pero explica brevemente términos técnicos y conserva un tono profesional y accesible.\n",
    "''',\n",
    "\"hard\": '''\n",
    "Reescribe el siguiente informe médico en español utilizando terminología médica precisa y estilo técnico, como si fuera dirigido a un lector profesional del ámbito sanitario. Conserva la complejidad del lenguaje, las estructuras formales y los matices clínicos, sin simplificar contenidos.\n",
    "'''\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "38131fad",
   "metadata": {},
   "outputs": [],
   "source": [
    "custom_promptsV1={\n",
    "\"easy\":'''\n",
    "Reescribe el siguiente informe médico en español con un nivel de lectura fácil correspondiente a un puntaje FH entre 70 y 100 (texto muy comprensible).\n",
    "Usa oraciones cortas y directas, vocabulario cotidiano, estructuras simples y explicaciones claras de términos médicos. El tono debe ser empático y accesible, como si estuvieras explicando la situación a un paciente o familiar sin conocimientos médicos.\n",
    "Mantén los datos clínicos y resultados esenciales, pero reemplaza o aclara tecnicismos con frases simples. Evita abreviaturas o siglas sin explicación.\n",
    "''',\n",
    "\"intermediate\": '''\n",
    "Reformula el siguiente informe médico en español con un nivel de lectura intermedio, correspondiente a un puntaje FH entre 50 y 70 (texto de dificultad moderada).\n",
    "Usa lenguaje formal pero comprensible, adecuado para lectores con educación general o estudiantes del área de salud. Mantén la precisión médica, pero agrega explicaciones breves tras los términos técnicos. Alterna oraciones simples y compuestas, con buena fluidez y cohesión.\n",
    "El texto debe sonar profesional, informativo y claro, sin llegar a la densidad típica de lenguaje técnico especializado.\n",
    "''',\n",
    "\"hard\": '''\n",
    "Reescribe el siguiente informe médico en español con un nivel de lectura avanzado o técnico, correspondiente a un puntaje FH entre 0 y 50 (texto especializado).\n",
    "Usa terminología médica precisa, estructuras sintácticas complejas y tono formal típico de documentos clínicos o publicaciones científicas. No simplifiques ni expliques los tecnicismos; conserva la exactitud conceptual y la nomenclatura profesional.\n",
    "Refleja el razonamiento clínico, hallazgos y juicios médicos con lenguaje apropiado para médicos, especialistas o investigadores.\n",
    "'''\n",
    "}"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "unsloth",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}