112 lines (111 with data), 3.7 kB

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The idea of this approach is to generate a prompt to ask a Pretrained LLM to label all the tokens with one of the labels we are considering."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "prompt_context = '''I need to perform a named entity recognition task on a  text related with inclusion criteria in clinical trials.\n",
    "The entities you need to recognize are: Condition, Value, Drug, Procedure, Measurement, Temporal, Observation, Person, Mood, Device and Pregnancy_considerations.\n",
    "Particularly you have to produce the ouput in the BIO format. I will show you an example of the expected output.\n",
    "Input text: Patients with symptomatic CNS metastases or leptomeningeal involvement \n",
    "Output:\n",
    "Patients O\n",
    "with O\n",
    "symptomatic O\n",
    "CNS B-Condition\n",
    "metastases I-Condition\n",
    "or O\n",
    "leptomeningeal B-Condition\n",
    "involvement I-Condition\n",
    "\n",
    "You can see that tokens without any entity are labeled as O, and the tokens that are part of an entity are labeled as B-<entity> or I-<entity> depending on if they are the beginning or the inside of the entity.'''\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def gen_prompt(text):\n",
    "    return f'''{prompt_context}\n",
    "Input text: {text}\n",
    "Given this input text, produce the output in the format I gave you.\n",
    "Output:'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "I need to perform a named entity recognition task on a  text related with inclusion criteria in clinical trials.\n",
      "The entities you need to recognize are: Condition, Value, Drug, Procedure, Measurement, Temporal, Observation, Person, Mood, Device and Pregnancy_considerations.\n",
      "Particularly you have to produce the ouput in the BIO format. I will show you an example of the expected output.\n",
      "Input text: Patients with symptomatic CNS metastases or leptomeningeal involvement \n",
      "Output:\n",
      "Patients O\n",
      "with O\n",
      "symptomatic O\n",
      "CNS B-Condition\n",
      "metastases I-Condition\n",
      "or O\n",
      "leptomeningeal B-Condition\n",
      "involvement I-Condition\n",
      "\n",
      "You can see that tokens without any entity are labeled as O, and the tokens that are part of an entity are labeled as B-<entity> or I-<entity> depending on if they are the beginning or the inside of the entity.\n",
      "Input text: Patients with symptomatic CNS metastases or leptomeningeal involvement\n",
      "Given this input text, produce the output in the format I gave you.\n",
      "Output:\n"
     ]
    }
   ],
   "source": [
    "# Example of a generated prompt\n",
    "p = gen_prompt('Patients with symptomatic CNS metastases or leptomeningeal involvement')\n",
    "print(p)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "TER",
   "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": "3.10.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}