Follow Datanami:
August 9, 2019

Big Data Yields Progress in Social Good

(Cienpies Design/Shutterstock)

In addition to driving profitability and growth for business, big data holds the potential to improve people’s lives in more humanitarian ways. One company that’s doing more than most in this arena is IBM, which last month provided an update on the Science for Social Good initiative that it started three years ago.

Statistics IBM shared by IBM indicate the Science for Social Good program is having an impact. Since the program started, 114 IBM Research scientists from seven global labs have contributed their time to 28 different data science endeavors. The work has involved 18 non-governmental organizations (NGOs) and resulted in 47 scientific papers and nine patents.

The data work has utilized machine learning and AI techniques to explore and find potential solutions to a variety of societal problems, including finding causes of opioid addiction, recognizing hate speech, detecting new epidemics and diseases, and spotting humanitarian crises in real time.

Aleksandra (Saška) Mojsilović, an IBM Fellow at IBM Research and a co-director of the campaign, says the endeavor’s goal is to create scalable technology solutions that can be reused to address other problems. IBM works with the United Nations to help identify areas that it can help.

(Source: “Interpretable Subgroup Discovery in Treatment Effect Estimation with Application to Opioid Prescribing Guidelines”)

“The idea was to kind of tap into our scientific talent to try and find novel technical solutions…. that are grounded in data and AI that could be used to address the most critical social and humanitarian challenges of the world,” Mojsilović says in a video that accompanied a blog post on the topic. “I think that’s really something really unique to IBM Research because I can think of very view places in the world that can offer that breadth and scale of expertise and passions.”

IBM has five main focus areas for 2019, including another project targeting the opioid epidemic. Considering that many cases of addiction start with a prescription, IBM Research scientists and experts from IBM Watson Health will team up will investigate how machine learning technologies and other methods can be used to create an early warning system.

The company is also partnering with Cincinnati, Ohio-based CityLink Center to investigate possible pathways out of poverty. This project will utilize a longitudinal dataset to detect a causal model of events and outcomes along several variables, such as employment, wellness, education, and housing.

Cures for particular types of cancer are few and far between. In a joint project with Cures Within Reach for Cancer, IBM scientists will attempt to find new cancer treatments among drugs that were not developed for that purpose. The program will utilize natural language processing (NLP) techniques to analyze large amounts of literature to find off-patent drugs that could be good candidates for further testing as cancer treatments.

(Source: “The Effect of Extremist Violence on Hateful Speech Online”)

A second cancer-related study is also in the works, this one having to do with skin cancer diagnoses. While lighter skinned people have the highest risk for developing skin cancer, the mortality rate among blacks in the US is much higher, due primarily to misdiagnosis. IBM Research will investigate machine learning methods for melanoma detection.

Finally, IBM researchers will investigate ways to make machine learning more transparent and knowable with the Trustworthy AI Pentathlon. Having machine learning models that are accurate isn’t good enough. They also must be explainable, fair, robust to data shift, and robust to adversarial examples, IBM says. The researchers will look to develop methods for achieving success across all five dimensions.

Big data and AI hold great potential to do both good and harm. With the right mindset, goals, and concerns for others, people can steer the technology to achieve the former and not the later. IBM’s Science for Good is a prime example of how corporations and individuals can work together to use technology to further humanity.

“We become better scientists and better engineers and better human beings,” Mojsilović says, “and I think that’s probably the main point off running IBM Science for Social Good.”

Related Items:

Ethical Data Science Is Good Data Science

The Humanity in Artificial Intelligence

Big Data, Big Problems? Responsible Data Management in 2019