Through the use of big data, the New York Blood Center hopes to attract the specific donors they need.
Could you make big data useful for blood donation? The New York Blood Center (NYBC) said yes—which is why it's now executing a big data strategy to reach the right blood donors, together with artificial intelligence and predictive analytics health tech startup Hindsait. The company applies artificial intelligence and predictive analytics to big data sets to deliver better healthcare. We interviewed co-founder and CEO Pinaki Dasgupta who told nuviun more about his work with the New York Blood Center and the future of big data in healthcare.
Starting February 1st, the city of New York began applying a machine learning approach to find and contact people who would be most likely to donate blood. According to an article published by the New Scientist, the New York Blood Center (NYBC) is using this approach to target and increase the turnout of African American donors.
What's the Problem?
Although blood is made of the same basic elements, certainly we know that all blood is not the same. Type B blood is more common in African Americans. The American Red Cross states:
If the donor and recipient are from the same ethnic background, the chance of a reaction can be reduced and that African-American blood donation may be the best hope for the needs of patients with sickle cell disease, 98% of whom are of African-American descent.
On a national basis, blood donation rates of African Americans are only between 25-50% compared to their white colleagues. A study from 2007 suggests three possible reasons for the discrepancy.
- increased rates of donor deferral and ineligibility
- increased barriers to donation, such as fear and distrust
- different marketing and education strategies
The study says that the issue needs to be better understood and thanks to New York Blood Center’s big data approach and we might see great progress.
Beth Shaz, who wrote the paper “Minority Donation in the United States: Challenges and Needs” (data from the article used above), is also chief medical officer at New York Blood Center. She said in an interview that minorities like African Americans provide blood that is needed for specific patient populations.
It’s critical to find and get the right kind of blood donors to the blood centers. By using the big data approach, the New York Blood Center leverages the power of personalization—including sending out individualized messages on various channels, while taking into account the particular metrics that can increase the chance of having more people respond.
How AI Technology is Applied to Big Blood Donor Data
Hindsait applies natural language processing to the data that helps to unlock secrets, and prepares the data for analysis and machine learning. Over 10 million records were analyzed, and by using artificial intelligence, Hindsait developed an updated HIPAA-compliant approach.
It allowed them to develop an algorithm to predict how to best contact donors. The data for the algorithm Hindsait used took into account metrics like:
- Last time a donor donated blood
- Eligibility for recent visits
- Age of donor
New York Blood Center was brave to push forward with a new big data approach, and it made perfect sense. It will be interesting to see how effective the campaign turns out to be. Although many algorithms don't work as well as some would hope (some big data algorithms for unstructured data, a study has found, have only 90% accuracy and 80% reproducibility)—we look forward to seeing what the outcome will be here.
From Politics to Healthcare
Obama’s “precision medicine,” first mentioned in his State of the Union address may have its shortcomings, but still managed to impress many in the US who didn’t expect that big data in medicine would make it on a large government’s agenda. Making use of big data is not new to Obama and his team. The administration was pioneering it as part of the election strategy and used machine learning to approach voters for the 2012 election campaign. Knowing who to contact, what to say, and how to do it to get them hooked was critical for its success—with a goal of finding better ways to target voters. And it worked.