Healthcare organization uses monitoring tool and language understanding to find blood donor search requests in social media
- Customer:
- DonorUA
- Industries:
- Healthcare, Nonprofit
- Tags:
- Natural Language Processing, AI for Good, Social Media Monitoring
- Technologies:
- Language Understanding
According to WHO, based on samples of 1000 people, the blood donation rate is 31.5 donations in high-income countries, 15.9 in upper-middle-income countries, 6.8 in lower-middle-income countries, and 5.0 in low-income countries.
Social networks, e.g., Facebook, Instagram, or Twitter, are full of posts from people, healthcare organizations, and blood centers asking to donate blood. Most of these posts remain ignored.
Urgent:
— ALI SHAN (SATTO) (@SatoMexicanox) August 11, 2021
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.
#Emergency 🚨
— Divya Sri ™ (@Divya__Official) August 11, 2021
Patient - Venu
Age - 26/M
Blood Group - B+ve Platelets SDP Donor
Units Needed - 1
Platelet count - 27K
Cause - Dengue case
Hospital - Shreyas Hospital
Location - Bhavanipuram , Vijayawada
Contact - 9948304069
ThankYou In Advance ❤🙏
Urgent:
— Maldives Blood Donors 🇲🇻❤️🇵🇸 (@MVBlooddonors) August 11, 2021
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.
#Delhi #Urgent Need O+/Any group #blood donor @ AIIMS hospital.
— BloodAid (@BloodAid) October 23, 2021
Call 9593778308#BloodAid via @SangitaGhalay @baxishweta cc @TajinderBagga @upma23 @ArvindGaur #BloodMatters
Having such blood donation requests could help local NGOs and organizations like Red Cross and World Health Organization to perform a real-time analysis in a specific city, country, or region and react fast. Historical data analysis can help to predict blood demand and supply and recruit donors before the shortage happens.
The solution
To archive this goal, we develop an Azure application to monitor posts in social media, extract meaningful information, and store data for future analysis. As a proof of concept, we use a social media monitoring tool from YouScan to find Twitter posts containing requests like "blood donors required" or "blood donors needed".
Also, we can get tweets related to blood donation that is not blood donor requests. So the model performs the classification job and detects if a publication is a blood donor request. If not, we skip the tweet.
To extract meaningful information, we develop a language understanding model using the Language Understanding service from Microsoft. It can identify the blood center's location, the number of units required, blood group type and rhesus, contact details, etc.
Combining this data with a database of blood donors, we generate alerts and send notifications to blood donors in a specific city or region.
You can try a demo here: https://ai.donor.ua.