DonorUA Improves Blood Donor Search Through Social Listening and NLP

According to WHO, based on samples of 1000 people, the blood donation rate is 32.6 donations in high-income countries, 15.1 donations in upper-middle-income countries, 8.1 donations in lower-middle-income countries and 4.4 donations in low-income countries.

This study aims to optimize the recruitment of blood donors by leveraging social media for DonorUA nonprofit organization. The real-time analysis of donation requests across various platforms can offer invaluable insights, enabling organizations like the Red Cross and WHO to respond promptly and efficiently within specific regions or cities. Moreover, the historical data accrued over time can facilitate predictive analysis to anticipate and mitigate potential shortages in blood supply.

Here is an example of a post from Twitter with a request for blood donation:

We have engineered an application on Microsoft Azure to meticulously monitor and analyze blood donation requests on social media. The initial phase of the project utilizes YouScan, a sophisticated social listening tool, to identify and extract relevant posts from Twitter. Posts are filtered based on specific keywords and phrases such as "blood donors required" and "blood donors needed".

Our system is equipped with a robust classification model that discerns actual donor requests from unrelated posts. Non-pertinent posts are systematically excluded from the dataset. Additionally, we have integrated the Language Understanding service from Microsoft to enhance the extraction of meaningful and precise information from the collected data.

This service facilitates the identification of key details such as the location of blood centers, the specific blood type and Rh factor required, the quantity of blood units needed and pertinent contact information.

Large language models allow better understanding of a context and perform named entity recognition. As an example, we can extract all needed information just by using prompt engineering.

Given the following tweet:

Urgent: O+ blood needed for a patient ( kid ) at AEH, Addu City. Plz contact 7847565 if you can donate or can find a donor for the kid. Plz share and help.

It can be transformed into named entities like:

This data can be easily extracted and analysed to perform quick and professional support and healthcare services.

This innovative approach aims to revolutionize blood donation recruitment strategies by harnessing the power of social media analytics, facilitating a more responsive and effective blood donation system. Through meticulous data analysis, our model aspires to bolster the efforts of NGOs and global health organizations in securing a consistent and reliable blood supply.