Classifying Comments on Social Media Related to Living Kidney Donation: Machine Learning Training and Validation...

Area of Science:

• Nephrology
• Public Health
• Data Science

Background:

• Living kidney donation is a vital source of organs, yet barriers like medical ineligibility and costs limit participation.
• Existing research on donation perceptions is limited, often excluding the general public's views.
• Understanding public perceptions is crucial for developing effective strategies to boost living kidney donation rates.

Purpose of the Study:

• To develop and implement a novel web-scraping and machine learning methodology for collecting and classifying public online comments on living kidney donation.
• To create a publicly accessible dataset of online comments related to living kidney donation for future research.

Main Methods:

• Utilized Python-based web-scraping tools to gather comments from diverse online platforms including news sites and social media.
• Developed manual classification guidelines for training data and employed deep learning models for large-scale comment classification.
• Achieved 84% accuracy in deep neural network testing, with a 63% actual accuracy after validation.

Main Results:

• Collected over 203,000 unique comments, with 11,027 identified as relevant to living kidney donation.
• The machine learning model demonstrated effectiveness in filtering and categorizing public discourse on the topic.
• The resulting dataset provides a rich resource for analyzing public sentiment and concerns regarding living kidney donation.

Conclusions:

• Web-scraping and machine learning offer a scalable and effective approach to gauge public opinion on living kidney donation.
• This methodology facilitates comprehensive analysis of perceptions, myths, and sentiments surrounding living kidney donation.
• The findings pave the way for data-driven interventions to address barriers and promote living kidney donation.