The opioid crisis depicted with a map of the USA filled with oxycodone and hydrocodone pills.
The largest ever seizure of fentanyl in the United States – 254 pounds of the white powder, enough to kill 1 in 3 Americans by overdose – was found under a shipment of cucumbers recently.
A policing approach alone is insufficient to take on the opioid crisis.
Those types of stories barely make the headlines any more, in part because illicit drugs are no longer just handsold by drug dealers; these sales have gone online. The neighborhood dealer faces the same evolving environment as other retailers and may soon go the way of Sears.
But opioids themselves are not going away. I could make an opioid purchase online in about 30 seconds and have it sent to my door, says Joe Smyser. The epidemiologist and president of The Public Good Projects isn't bragging, he's simply stating a fact about the opioid crisis that has struck the United States. The U.S Drug Enforcement Agency, social media companies, and some foreign governments have undertaken massive efforts to shut down sites selling illegal drugs, and they have gotten very good at it, shuttering most within a day of their opening.
But it's a Whac-A-Mole situation in which new ones pop up as quickly as older ones are closed; they are promoted through hashtags, social media networks, and ubiquitous email spam to lure visitors to a website or call a WhatsApp number to make a purchase. The online disruption by law enforcement has become simply another cost of doing business for drug sellers. Fentanyl, and similar analogues created to evade detection and the law, are at the center of it. Small amounts can be mixed with other "safer" opioids to get a high, and the growth of online sales have all contributed to the surge of opioid-related deaths: about 17,500 in 2006; 47,600 in 2017; and a projected 82,000 a year by 2025.
All of this has occurred even while authorities have been cracking down on the prescribing of opioids, and prescription-related deaths have declined. Clearly a policing approach alone is insufficient to take on the opioid crisis.
Building the Tools
The Public Good Projects (PGP), a nonprofit organization founded by concerned experts, was set up to better understand public health issues in this new online environment and better shape responses. The first step is to understand what people are hearing and the language they are using by monitoring social media and other forms of public communications. "We're collecting data from every publicly available media source that we can get our hands on. It's broadcast television data, it's radio, it's print newspapers and magazines. And then it's online data; it's online video, social media, blogs, websites," Smyser explains.
The purpose was to better understand the opioid crisis and find out if there were differences between affected rural and urban populations.
"Then our job is to create queries, create searches of all of that data so that we find what is the information that Americans are exposed to about a topic, and then what … Americans [are] sharing amongst themselves about that same topic."
He says it's the same thing business has been doing for years to monitor their "brand health" and be prepared for possible negative issues that might arise about their products and services. He believes PGP is the first group to use those tools for public health.
Looking At Opioids
PGP's work on opioids started with a contract from the Substance Abuse and Mental Health Administration (SAMHSA) through the National Science Foundation. The purpose was simply to better understand the opioid crisis in the United States and in particular find out if there were differences between affected rural and urban populations. A team of data scientists, public health professionals, and cultural anthropologists needed several months to sort out and organize the algorithms from the sheer volume of data.
Drug use is particularly rich in slang, where a specific drug or way of using it can be referred to in multiple ways in different towns and social groups. Traditional media often uses clinical terms, Twitter shorthand, and all of that has to be structured and integrated "so that it isn't just spitting out data that is gobbledygook and of no use to anyone," says Smyser.
The data they gather is both cumulative and in real time, tabulated and visually represented in constantly morphing hashtag and word clouds where the color and size of the word indicates the source and volume of its use.
Popular hashtags on Twitter relating to the opioid crisis.
(Credit: The Public Good Projects)
The visual presentation of data helps to understand what different groups are saying and how they are saying it. For example, compare the hashtag and word clouds. Younger people are more likely to use the hashtags of Twitter, while older people are more likely to use older forms of media, and that is reflected in their concerns and language in those clouds.
Popular words relating to the opioid crisis gathered from older forms of media.
(Credit: The Public Good Projects)
A Ping map shows the origin of messages, while a Spidey map shows the network of how messages are being forwarded and shared among people. These sets of data can be overlaid with zip code, census, and socioeconomic data to provide an even deeper sense of who is saying what. And when integrated together, they provide clues to topics and language that might best engage people in each niche.
A Ping map showing the origin of messages around the opioid crisis.
(Credit: The Public Good Projects)
Opioids Speak
One thing that quickly became apparent to PGP in monitoring the media is that "over half of the information that the American public is exposed to about opioids is a very distant policy debate," says Smyser.
It is political pronouncements in DC, the legal system going after pharmaceutical companies that promoted prescription opioids for pain relief (and more), or mandatory prison terms for offenders. Relatively little is about treatment, the impact on families and communities, and what people can do themselves. That is particularly important in light of another key finding: residents of "Trump-land," the rural areas that supported the president and are being ravaged by opioids, talk about the problem and solutions very differently from urban areas.
"In rural communities there is usually a huge emphasis on self-reliance, and we take care of each other; that's why we enjoy living here. We are a neighborhood, we come together and we fix our own problems," according to Smyser.
In contrast, urban communities tend to be more transient, less likely to live in multigenerational households and neighborhoods, and look to formal institutions rather than themselves for solutions. "The message that we're sending people is one where there is really no role whatsoever for self-efficacy...we're giving them nothing to do" to help solve the problem themselves, says Smyser. "In fact, I could argue it is reducing self-efficacy."
Residents of "Trump-land," the rural areas that supported the president and are being ravaged by opioids, talk about the problem and solutions very differently from urban areas.
The opioid crisis is complex and improving the situation will be too. Smyser believes a top-down policing approach alone will not work; it is better to provide front-line public health officers at the state and local level with more and current intelligence so they can respond in their communities.
"I think that would be enormously impactful. But right now, we just don't have that service." SAMHSA declined multiple requests to discuss this project paid for with federal money. A spokesman concluded with: "That project occurred under the previous administration, and we did not have a direct relationship with PGP. As a result, I am unable to comment on the project."
The Milken Institute Center for Public Health, a think tank that is working to find solutions to the opioid epidemic, had an upbeat response. Director Sabrina Spitaletta said, "PGP's work to provide real-time data that monitors topics of high concern in public health has been very helpful to many of the front-line organizations working to combat this crisis."
A recent study in The Lancet Oncology showed that AI found 20 percent more cancers on mammogram screens than radiologists alone.
Published in The Lancet Oncology, the study analyzed the scans of 80,000 Swedish women with a moderate hereditary risk of breast cancer who had undergone a mammogram between April 2021 and July 2022. Half of these scans were read by AI and then a radiologist to double-check the findings. The second group of scans was read by two researchers without the help of AI. (Currently, the standard of care across Europe is to have two radiologists analyze a scan before diagnosing a patient with breast cancer.)
The study showed that the AI group detected cancer in 6 out of every 1,000 scans, while the radiologists detected cancer in 5 per 1,000 scans. In other words, AI found 20 percent more cancers than the highly-trained radiologists.
Scientists have been using MRI images (like the ones pictured here) to train artificial intelligence to detect cancers earlier and with more accuracy. Here, MIT's AI system, MIRAI, looks for patterns in a patient's mammograms to detect breast cancer earlier than ever before. news.mit.edu
But even though the AI was better able to pinpoint cancer on an image, it doesn’t mean radiologists will soon be out of a job. Dr. Laura Heacock, a breast radiologist at NYU, said in an interview with CNN that radiologists do much more than simply screening mammograms, and that even well-trained technology can make errors. “These tools work best when paired with highly-trained radiologists who make the final call on your mammogram. Think of it as a tool like a stethoscope for a cardiologist.”
AI is still an emerging technology, but more and more doctors are using them to detect different cancers. For example, researchers at MIT have developed a program called MIRAI, which looks at patterns in patient mammograms across a series of scans and uses an algorithm to model a patient's risk of developing breast cancer over time. The program was "trained" with more than 200,000 breast imaging scans from Massachusetts General Hospital and has been tested on over 100,000 women in different hospitals across the world. According to MIT, MIRAI "has been shown to be more accurate in predicting the risk for developing breast cancer in the short term (over a 3-year period) compared to traditional tools." It has also been able to detect breast cancer up to five years before a patient receives a diagnosis.
The challenges for cancer-detecting AI tools now is not just accuracy. AI tools are also being challenged to perform consistently well across different ages, races, and breast density profiles, particularly given the increased risks that different women face. For example, Black women are 42 percent more likely than white women to die from breast cancer, despite having nearly the same rates of breast cancer as white women. Recently, an FDA-approved AI device for screening breast cancer has come under fire for wrongly detecting cancer in Black patients significantly more often than white patients.
As AI technology improves, radiologists will be able to accurately scan a more diverse set of patients at a larger volume than ever before, potentially saving more lives than ever.