Project Migraine: Part 2

💊Migraine Medications Considered

Commonly Used Migraine Preventive Medications

Medication Group	Generic Medication (US Brand Name)
Antiseizure medications	Topiramate (Topamax)
Beta Blockers	Propranolol (Inderal), Atenolol (Tenormin), Metoprolol (Toprol)
Tricyclic antidepressants	Amitriptyline (Elavil), Nortriptyline (Pamelor)
OnabotulinumtoxinA (Botox)	OnabotulinumtoxinA (Botox)
CGRP monoclonal antibodies	erenumab (Aimovig), galcanezumab (Emgality), fremanezumab (Ajovy), eptinezumab (Vyepti)
Gepants	Atogepant (Qulipta), rimegepant (Nurtec)

Commonly Used Migraine Acute Medications

Medication Group	Generic Medication (US Brand Name)
Triptans	Sumatriptan (Imitrex), Rizatriptan (Maxalt), Eletriptan (Relpax), Naratriptan (Amerge), Frovatriptan (Frova), Zolmitriptan (Zomig), Almotriptan (Axert)
Gepants	ubrogepant (Ubrelvy), rimegepant (Nurtec), zavegepant (Zavzpret)
Ditan	Lasmiditan (Reyvow)
Ergots	Dihydroergotamine (DHE, Migranal, Trudhesa), ergotamine (Cafergot)

🎯Tested Environment

The project has been tested on the following configurations:

• Operating System: Ubuntu 20.04.5 LTS
• NVIDIA-SMI 525.60.13
• Driver Version: 525.60.13
• CUDA Version: 12.0

👩‍💻Installing Conda environment:

1. Activate Conda in bash
2. Install packages using following command:

$ conda env create -f installations/environment.yml

3. If you only want to install packages without version: conda create --name <env-name> --file installations/packages.txt
4. Install any remaining packages using pip as follows:

$ pip install -r installations/requirements.txt

🏃Running the Classification Code

1. Clone the repositry: git clone https://github.com/swati-rajwal/migraine.git
2. Get access to dataset, not publicly shared at the moment. Alternatively, you can use this pipeline to your own dataset as well. Put the dataset in data folder
3. A_Dataset.py file is written specifically for the dataset we have. If you have access to our dataset, this file should work. Else, you will have to pre-process your data accordingly.
4. Create dataset splits: python B_nfold_split.py <csv_file_path> <output_folder_path>
5. Run chmod +x C_1_run_cls_multiGPU.sh to ensure you have rights to run this file.
6. Run C_1_run_cls_multiGPU.sh that in turn runs the C_2_simpletransformers_cls.py file for RoBERTa based classification
7. As an example, you can run a command like ./C_1_run_cls_multiGPU.sh &> results/roberta_run_$(date +%Y-%m-%d).log
8. If you re-run step 4, make sure to either delete or rename the folder 'model'
9. For evaluation, run python D_eval_model.py task_configs/migraine.json &> results/eval_run_$(date +%Y-%m-%d).log command.
10. To understand the sentiment across various medication groups, run python E_sentiment_analysis.py